et discipline ou spécialité
Jury :
le
Université Toulouse 3 Paul Sabatier (UT3 Paul Sabatier)
Keoni Saint-Pé
mardi 3 décembre 2019
Quantification in situ des mouvements de truite fario
ED SEVAB : Écologie, biodiversité et évolution
UMR5321
Nuria Sanz (Rapportrice)Jean Marc Roussel (Rapporteur)Arnaud Caudron (Examinateur)Thierry Oberdorff (Examinateur)Franck Cattaneo (Examinateur)
SImon Blanchet
ii
Résumé Les mouvements sont un aspect crucial de l‘écologie et de l‘évolution, car ils déterminent la dynamique des
populations et des communautés. Les menaces anthropiques pesant sur ces dynamiques sont donc une
préoccupation majeure pour la conservation. Dans cette thèse, j‘ai étudié les mouvements de la truite
commune (Salmo trutta) dans le contexte des rempoissonnements, c‘est-à-dire la supplémentation des
populations sauvages à partir de souches d‘élevage locales et exogènes.
À cette fin, j'ai d'abord développé un nouvel outil génétique utile pour détecter la structure génétique chez
la truite commune et l'hybridation avec des souches élevées en captivité. Cet outil a permis un grand succès
de génotypage et a permis d'identifier les patrons d'isolement par la distance. Son intérêt réside dans sa
facilité de mise en œuvre, et son universalité potentielle pour la génétique des populations de cette espèce
dans son aire de répartition. J‘ai ensuite développé de nouvelles méthodes d'assignation combinant des
données génétiques et isotopiques, afin d'étudier les mouvements à l'échelle spatiale d'un bassin
hydrographique, ce qui présente un intérêt pour des questions appliquées telles que la gestion. Cette
approche, basée sur le machine learning, a révélé une grande précision et un fort pouvoir discriminant pour
assigner des individus à leur population d‘origine. J‘ai aussi décrit les effets des rempoissonnements sur les
patrons de diversité et de différenciation génétiques et constaté que les rempoissonnements ont pour effet
d‘augmenter la diversité et la différenciation, et que les patrons naturels attendus pouvaient être inversés
dans le cas de cette pratique. Ensuite, l‘hybridation entre souches sauvages et d‘élevage affecte les patrons
de dispersion, révélant que les deux souches diffèrent en termes de propension, de distances et de direction
dans leurs dispersion, alors que les hybrides présentent des patrons de dispersion moins prononcés. Enfin,
j‘ai cherché à mieux comprendre l‘influence des facteurs individuels, environnementaux et paysagers sur
les mouvements des populations naturelles; ici, j‘ai trouvé que certains déterminants étaient universels d‘un
bassin versant à l‘autre, par exemple le fait que les gros individus étant plus enclins à se déplacer, ou encore
que les sites directement reliés par le flux d‘eau, et ceux ayant des similitudes en termes d‘altitude et de
type de cours d‘eau échangeaient plus de migrant que les autres. D‘autre part, d‘autres facteurs dépendaient
du contexte, par exemple, les relations entre patrons de mouvements et position dans les paysages fluviaux
et la disponibilité de l‘habitat varient entre les rivières.
Cette thèse a contribué à améliorer les méthodes d'étude des mouvements et à identifier les facteurs sous-
jacents aux patrons de mouvements à l'échelle du bassin hydrographique. Les implications de ma thèse sont
donc à la fois fondamentales et appliquées, car une meilleure compréhension des patrons de mouvement
dans le contexte de perturbations humaines est cruciale pour la gestion et la conservation.
iii
Abstract
Movements are a crucial aspect of ecology and evolution, as they determine population and
community dynamics. Threats to these dynamics because of human perturbations are therefore a major
concern for conservation. In this thesis, I studied movements in the Brown trout (Salmo trutta) in the
context of stocking, i.e. the supplementation of wild populations with captive-bred strains from both,
native and exogenous origin.
For this purpose, I first developed a new genetic tool useful for detecting genetic structure in the brown
trout, as well as hybridization with captive-bred strains, exhibiting high genotyping success and
enabling to successfully identify patterns of isolation-by-distance. This tool was shown cost effective,
and especially, should be useful for many population genetics studies on this species across its range.
Then, I developed novel assignment approaches combining genetic data and stable isotopes, to study
movements at the spatial scale of a river basin, which is of interest for applied matters such as
management. This approach, based on machine learning, revealed high accuracy and power to
discriminate and assign individuals to their population of origin. Further, I described the genetic effects
of captive breeding on patterns of genetic diversity and differentiation, and found that captive-bred
genotypes increased diversity and differentiation, and that expected natural patterns could be reversed
in the case of higher frequency of captive-bred genotypes occurring at the level of populations. Then, I
demonstrated that admixture between wild individuals and those carrying captive-bred ancestry
affected dispersal patterns, that the two strains displayed different movement patterns in terms of
propensity, distances, and direction, and that admixture between strains considerably reduced
dispersal. Finally, I aimed at better understanding how individual, environmental and landscape related
factors influence movements in natural populations; here I found that some determinants were
universal across rivers, with larger individuals being more prone to movement for instance, or sites that
are directly connected by the water flow, and those that are similar in terms of elevation and stream
order exchanged more migrants. One the other hand, other drivers were context dependent, for instance
the relations between movement patterns and position within riverscapes and habitat availability
depended on the river basin considered.
This thesis contributed to improve methods for studying movements, and to identify factors underlying
patterns of movements at the scale of the river basin. The implications of my thesis are thus both
iv
fundamental and applied as a better understanding of movement patterns in the context of human
perturbations is crucial for management and conservation.
Avant-Propos
Cette thèse a été réalisée au sein de la Station d‘Ecologie Théorique et Expérimentale du CNRS à
Moulis. Elle a été financée par un contrat de recherche engagé par l‘Agence Française pour la
Biodiversité (AFB; ex ONEMA), et Électricité de France (EDF).
Cette thèse est présentée en anglais, sous la forme d‘une introduction générale, de cinq chapitres
(sous forme d‘articles scientifiques) et d‘une discussion générale.
Le chapitre IV est publié dans Conservation Genetics, le chapitre I est publié dans BMC
Genomics. Les chapitres II, III et V sont en préparation pour soumission dans des revues avec
comité de lecture.
vi
Remerciements
Je tiens tout d‘abord à remercier les cinq personnes qui ont encadré ma thèse : Simon Blanchet,
Laurence Tissot, Nicolas Poulet, Jérôme Prunier et Julien Cucherousset. Merci à vous pour tout le
soutien que vous m‘avez apporté. Simon, merci pour ces cinq années à tes côtés, depuis mon stage
de Master 1. Ça a été un véritable plaisir de travailler avec toi. Tu as été le directeur de thèse
parfait en sachant à la fois me laisser des libertés dans mon travail, et me mettre la pression aux
bons moments pour que ma thèse avance, sans pour autant m‘empêcher d‘aller à la pêche. Jérôme,
merci pour ta pédagogie, ton dévouement et ta patience qui ont été d‘une aide précieuse pour mes
travaux. Mais aussi, je vous remercie pour les moments hors cadre professionnel : parties de
pêche, chasse à l‘ours, coinches et tout le reste ont aussi fait de mes années de thèse des moments
exceptionnels.Julien, merci pour ton soutien scientifique sur la truite et l'isotopie, qui a permis de
soulever des choses importantes et de bien diriger mes analyses. Laurence et Nicolas, merci pour
m‘avoir fait confiance dès le début, et pour avoir permis que cette thèse voie le jour. Merci
égalemet pour votre soutien permanent, et votre aide au cours de mes travaux.
Je tiens à remercier également mes deux rapporteurs : Nuria Sanz, Jean-Marc Roussel, pour avoir
accepté avec intérêt de prendre le temps de lire et corriger mon manuscrit pendant l‘été ainsi que
de faire le déplacement pour assister à ma soutenance.
Je souhaite aussi remercier l‘équipe de Montpellier : Maeva Leitwein, Pierre-Alexandre Gagnaire,
Patrick Berrebi et Bruno Guinand, qui m‘ont permis d‘utiliser leurs travaux, et sans qui ma thèse
aurait été très différente. Un grand merci aussi aux membres de mon comité doctoral Pierre-
Alexandre, Patricia Balaresque et mes encadrants, qui m‘ont conseillé sur mes choix et stratégies.
Merci à toutes les personnes qui ont participé au terrain, Geoffrey, Morgane, Pierre, Charlotte, et
aussi les membres des AAPPMA, en particulier celle de Luchon : Jean Lérimé et Daniel Estrade et
les autres membres ainsi que celle de Saugues, Jean-Claude Pulveric et ses collègues, qui rendent
ces associations très dynamique en faisant participer le monde scientifique.
Un grand merci aussi à mes colocataires de la maison de Luzenac: Jeff, Matthieu, Jonathan,
Azenor, sans oublier Poncho, merci pour votre bonne humeur, votre soutien, et pour tous les bons
vii
moments passé ensemble, même en hiver sans chauffage! Merci également à tous mes collègues,
Jérémie, Kevin, Allan, Vinicius, Eloïse, Camille, Hugo, Rik, Lauranne, Julien, Félix, Thibault,
Nicolas, Paul, Louis, Alice, Léa, Lisa, Elise, Eglantine, Olivier, Diego, Nuria, Yuval, Arielle, et
tous ceux que j‘oublie pour les moments passés à Moulis ou ailleurs!
Enfin, je remercie très profondément ma famille, avec une pensée toute particulière pour mon
grand-père qui m‘a transmis sa passion, sans laquelle je n‘aurais jamais fait une thèse sur la truite!
Merci aussi à toute ma famille, ainsi qu‘à mes amis pour leur soutien dans mes projets. Pour finir,
mes derniers remerciements iront à Aisha et Carlos, votre présence et soutien affectif tout au long
de ma thèse ont été décisifs.
8
Table des matières
Résumé.............................................................................................................................................. ii Abstract ........................................................................................................................................... iii Avant-Propos ................................................................................................................................... v Remerciements ................................................................................................................................ vi
Table des matières ........................................................................................................................... 8 Introduction ................................................................................................................................... 10
Context ........................................................................................................................................ 10 Movements: concepts and definitions ......................................................................................... 11 Determinants and consequences of animal movement ................................................................ 12
Connectivity: a crucial characteristic of landscapes for all types of movements ........................ 16 Detecting/measuring movements ................................................................................................ 19 The brown trout, a fascinating species to study movements using molecular markers .............. 28
Stocking practices in europe, and effect on movement inference ............................................... 30 Movement patterns in brown trout populations........................................................................... 31
Aims of the phd ........................................................................................................................... 37
Chapter I. Development of a large SNPs resource and a low-density SNP array for brown
trout (Salmo trutta) population genetics ...................................................................................... 40 Abstract........................................................................................................................................ 41 Background.................................................................................................................................. 42
Methods ....................................................................................................................................... 44 Results ......................................................................................................................................... 49
Discussion.................................................................................................................................... 55 Conclusion ................................................................................................................................... 60
Declarations ................................................................................................................................. 60 References ................................................................................................................................... 61 Supplementary material for Chapter I ......................................................................................... 68
Chapter II. Combining genetics, stable isotopes and morphometrics improves assignments
for movement detection: a case study in Brown trout (Salmo trutta) populations .................. 74 Summary...................................................................................................................................... 75
Introduction ................................................................................................................................. 76 Methods ....................................................................................................................................... 80 Results ......................................................................................................................................... 85 Discussion.................................................................................................................................... 88 Conclusions ................................................................................................................................. 92
References ................................................................................................................................... 93 Supplementary material for Chapter II ........................................................................................ 97
Chapter III. Captive-bred ancestry affects spatial patterns of genetic diversity and
differentiation in Brown trout (Salmo trutta) populations ....................................................... 102 Summary.................................................................................................................................... 103 Introduction ............................................................................................................................... 104 Methods ..................................................................................................................................... 106
Discussion.................................................................................................................................. 118 Conclusions and implications .................................................................................................... 122
9
Authors' contributions ............................................................................................................... 122 Acknowledgments ..................................................................................................................... 123 Funding ...................................................................................................................................... 123 Conflict of Interest Statement .................................................................................................... 123
References ................................................................................................................................. 123 Supplementary material for Chapter III. ................................................................................... 130
Chapter IV. Genetic admixture between captive-bred and wild individuals affects patterns of
dispersal in a brown trout (Salmo trutta) population ............................................................... 138 Summary.................................................................................................................................... 139
Introduction ............................................................................................................................... 140 Material and Methods ................................................................................................................ 142 Results ....................................................................................................................................... 147
Discussion.................................................................................................................................. 151 Data accessibility ....................................................................................................................... 155 Authors' contributions ............................................................................................................... 155
Acknowledgments ..................................................................................................................... 155 Funding ...................................................................................................................................... 155
Conflict of Interest Statement .................................................................................................... 155 References ................................................................................................................................. 156 Supplementary material for Chapter IV. ................................................................................... 161
Chapter V. Patterns of brown trout (Salmo trutta) movements and underlying drivers in
replicated riverscapes: an approach combining isotopic and genetic markers ..................... 174 Summary.................................................................................................................................... 175 Introduction ............................................................................................................................... 176
Methods ..................................................................................................................................... 179 Results ....................................................................................................................................... 185
Discussion.................................................................................................................................. 192 Acknowledgments ..................................................................................................................... 196 Funding ...................................................................................................................................... 196
References ................................................................................................................................. 196 Supplementary material for Chapter V. ..................................................................................... 202
General discussion ....................................................................................................................... 210 Factors influencing movement in the brown trout, and management/conservation implications
................................................................................................................................................... 211 Studying movement in a managed and stocked species ............................................................ 215 Insights of detecting movements with individual based methods ............................................. 217 Conclusion and perspectives ..................................................................................................... 222
References .................................................................................................................................... 224
10
Introduction
CONTEXT
“The ability of individuals to move in space, although highly variable between species, is a
general characteristic of all organisms. In fact the very persistence of species on a micro-
evolutionary time-scale requires the ability to escape from natural, long-term environmental
changes. At shorter time-scales, movement capacities of organisms are universally important to
ecological phenomena. For this reason, the study of movement patterns deserves a key position
within all disciplines of ecology‖ (Ims, 1995)
Studying movements is a challenging yet crucial aspect of life history and behavior, thus of
population and community ecology, evolution, but also conservation (Hobson, 1999; Wocher and
Rösch, 2006). Movements are key to species persistence and coexistence at various spatial and
temporal scales (Ciancio et al., 2008; Hobson and Wassenaar, 1997; Kneitel and Chase, 2004) as
they are directly linked to the dynamics of populations (Bowler and Benton, 2005; Clobert, 2001;
Kokko, 2006), enabling to reduce risk of extinction for instance (Hanski, 1998). In the last 4 years,
I have studied movements of the Brown trout (Salmo trutta) in the context of human induced
perturbations, which is the general topic of my PhD thesis. Movements have been studied widely
in this iconic species, but generally concern either very local scales (a few hundreds of meters
during seasonal migration or dispersal of fry for instance; Acolas et al., 2008; Bagliniere et al.,
1989; Bembo et al., 1993; Cuinat and Heland, 1979; Dieterman and Hoxmeier, 2011; Gowan and
Fausch, 1996; Klemetsen et al., 2003; Maisse and Bagliniere, 1990; Ovidio et al., 1998), or very
large spawning or feeding movement with a majority of studies on migration between freshwater
and saltwater habitats (McCarthy and Waldron, 2000). Moreover, these studies have been done
mainly using direct estimates of movement (mark-recapture, PIT tagging), genetics, or otoliths
isotope analysis. Thus the novelties in my works –and which constitute the main objectives of my
thesis- consisted in developing non-lethal indirect methods: a new genetic tool, and novel
approaches combining genetic data as well as fin tissue stable isotopes, to study movements in the
context of anthropic perturbations, in particular stocking, at spatial scales which are of interest for
river managers, i.e. at scales of a river basin (~ few kilometers). Before going deeper in these
11
objectives, I will first define what the main types of movements consist in, and specify which
aspects of movement biology are concerned by my PhD works.
MOVEMENTS: CONCEPTS AND DEFINITIONS
Movements and connectivity are very complex concepts, and their definitions are highly important
to accurately study them. A movement is a spatial displacement of an individual (Ims, 1995),
which takes place in three steps: departure, transience and settlement. A movement can therefore
have various biological causes and consequences, and take many different forms. Although there
is no real consensus, because the type of movement and its definition can highly vary depending
on the questions of a study (Table 1; Koehn and Crook, 2013), animal movements can be
separated into four different categories: prospecting, foraging, migration and dispersal. These
different mechanisms each have different causes and consequences (Ims, 1995), and, depending on
the populations, spatial/time scales and species considered, they can be very difficult to
disentangle. I hereafter provide a brief definition of each of these movement types:
1) Prospecting movements are defined as visits made by individuals to locations where they are
not currently breeding, but where they may (or not) settle in the future to reproduce (Reed et al.,
1999).
2) Foraging movements consist in movements from home habitat to feeding habitat. These
movements usually occur at relatively small spatial and temporal scales, although this is not
always true (Péron and Grémillet, 2013), and are also known as trivial or daily movements
(Johnson, 1969).
3) Migrations consist in to-and-from movement of populations (or a large part of the population)
between regions where conditions are alternately favorable or unfavorable, and usually including
one region in which breeding occurs (Dingle and Drake, 2007), which is therefore temporally
predictable (Mehner, 2012). Migration differs from foraging and prospection because it occurs at a
much greater scale, and involves movement of much longer duration than those of normal daily
activities (Dingle and Drake, 2007).
4) Dispersal is defined as the movement of individuals between a location where they were born
or bred to a new location where they breed. Technically, dispersal is defined as any movement that
12
has the potential to lead to gene flow (Clobert, 2001; Ronce, 2007). Dispersal differs from
foraging or migration movements as it implies relocation of the natal/breeding site. Another
definition of dispersal (literal definition; Begon et al., 1988) would be a movement of individuals
which decreases their concentration (and increases the spatial range), for instance movement of
fishes from drought refuge after the drought is over. Other definitions exist, for instance a broader
one would be movements leading to redistribution within a spatially extended population (Dingle
and Drake, 2007; Gatehouse and Woodrow, 1987; Taylor, 1986).
Movement scale Description Purpose
Micro-scale Short, non-sustained Escape from predators, prey capture
Meso-scale Short term but sustained, within
normal home range
Feeding, avoidance of local poor conditions,
daily movements, spawning
Macro-scale
Prolonged, long-term, large
scale
movements
Migrations, exploration, feeding, spawning,
avoidance of larger scale
poor conditions, dispersal
Table I1: Different scales of movement (Koehn and Crook, 2013).
These different types of movements have been a vast field of research, in many species, and many
studies have aimed at understanding both their causes, and their consequences. Here I did not
focus on migratory patterns, but rather on small-scale to medium-scale (within rivers and between
main sections of a river network) movements that mainly encompass exploratory movements and
dispersal per se, but migrations can also be part of the movements detected here although it is
difficult to be certain that they are disentangled.
DETERMINANTS AND CONSEQUENCES OF ANIMAL MOVEMENT
Although some movements are random, and unlinked to an individual‘s or species‘ strategy,
movement patterns were widely shown to have strongly been selected for at the individual level
(Edelaar and Bolnick, 2012), but also at the population and species level (Shaw and Couzin,
2013). Indeed, species have evolved very particular movement patterns, which are part of their life
history, and thus crucial for their persistence (Jones, 1977; Nathan et al., 2008). Movements
therefore depend on a wide variety of individual and environmental features (Johnson et al., 2002).
13
Foraging movements and exploration
Food resources are often distributed heterogeneously, and the ability to efficiently exploit them is
under natural selection (Barton et al., 1992; Charnov, 1976). Foraging movements thus exhibit a
very wide range of strategies and features (O‘Brian et al., 1990). These movements have been
widely studied, and linked to individual differences in morphology (Armstrong et al., 1997;
Riveros and Gronenberg, 2010), behavior (Bacon et al., 2010; Caraco, 1981; Caraco et al., 1990),
sex; (Matsumoto et al., 2017), cognitive ability/memory (Rakow et al., 2010), and to
environmental features such as resource distribution (Grevstad and Klepetka, 1992; Mercier and
Lenoir, 1999), predation (Sweeney et al., 2013), and social interactions (Mehlhorn et al., 2015;
Patrick et al., 2017, 2013).
Migration
Migratory movements exist at a wide range of spatial and temporal scales (Lucas et al., 2001).
Most seasonal migrations have a determinant function in completing the life cycle of species
(Runge et al., 2014), because habitat use differs along species‘ cycle, therefore, each cycle may
require a certain habitat type (Aarestrup et al., 2009; Brönmark et al., 2008). Although mainly
linked to reproduction (Rakowitz et al., 2008), migrations can also consist in refuge migrations,
for instance against predation (Brönmark et al., 2008; Romare and Hansson, 2003; Skov et al.,
2013), or adverse environmental conditions (Berthold and Helbig, 2008). Finally, feeding
migrations are also observed in many species (Erman, 1981; Rikardsen et al., 2008). Forces that
shape the evolution of migratory behavior in animals are thus diverse (Brodersen et al., 2008;
Chapman et al., 2011; Olsson et al., 2006; Skov et al., 2013), and migration has shaped physiology
and life-history of species (Brönmark et al., 2014). Migration generally concerns a fraction of
individuals/populations, therefore, the causes and consequences of migration propensity have been
widely studied (Chapman et al., 2011). At the individual level, migration can increase survival
(Skov et al., 2013), however its cost is not always affordable and thus sometimes selected against
(Plath et al., 2013). For instance, individuals exhibiting poor physiological condition may be
forced to adopt a resident strategy (Brodersen et al., 2008). At the population level, intraspecific
competition was shown to be a driver of avian migration patterns for instance (Lundberg, 1987).
Migration was also widely shown to be environmentally driven (Pulido and Berthold, 2003; van
14
Noordwijk et al., 2006). Migration can strongly influence both trophic dynamics (Brönmark et al.,
2014; Post et al., 2008), for instance on the persistence of plant-herbivore systems (Fryxell and
Sinclair, 1988), but also in shaping population structure (Rolshausen et al., 2013; Wood and Foote,
1996).
Dispersal
Although random dispersal is a trait selected for species, particularly plants (Hamrick and
Loveless, 1986), but also animals (Byers, 2001), dispersal has been widely studied as individual,
environment and population driven (Barton et al., 2009; Clobert, 2001; Cote et al., 2010), and
depending on the stage of dispersal (Figure 1). At the individual level, one can for instance
increase its fitness by dispersing out of changing environments (Jackson and Sax, 2010), or to
avoid competition for resource or mates (Gowaty, 1993; Negro et al., 1997). This increase in
fitness has thus led evolution to shape some individuals to have improved dispersal traits (Cote et
al., 2010; Lowe and McPeek, 2014), also benefiting the species. These traits are diverse, and can
be phenotypical or behavioral (Clobert et al., 2009; Cote et al., 2010; Dingemanse et al., 2003).
Environments are also crucial in shaping dispersal patterns such as source-sink dynamics. In river
ecosystems for instance, the shape of the river network and the water flow create very particular
dispersal patterns (downstream sinks due to higher connectivity and downstream water-flow
directed dispersal; Paz-Vinas et al., 2015). Finally, another mechanism driving dispersal consists
in demo-genetic features such as population density (McPeek and Holt, 1992), kin competition and
inbreeding risk avoidance (Branch, 1975; Szulkin and Sheldon, 2008), leading to patterns such as
sex biased dispersal for instance (Greenwood, 1980; Li and Kokko, 2019).
15
Figure I1: Relation between the 3 stages of dispersal, phenotype and context (Clobert et al., 2009).
The biology of dispersal is therefore a fundament for many areas of ecology and evolutionary
biology (Bowler and Benton, 2005; Clobert, 2001; Kokko, 2006), as it has a direct link to
population dynamics (Figure 2). Indeed, dispersal is a key process for populations as it reduces
risk of extinction through counterbalancing genetic drift for instance, and many studies show that
impeding dispersal through fragmentation for instance can rapidly lead to collapse of a
population/species ―...dispersal is probably the most important life history trait involved in
both species persistence and evolution...‖ (Clobert, 2001)
Figure I2: Evolutionary (A) and ecological (B) loop affecting dispersal (Bowler and Benton 2004)
16
Links between types of movements
Interestingly, all types of movement are closely linked. Although dispersal movements were
shown to be clearly distinguishable from foraging/exploratory movements in some cases (Zera and
Denno, 1997), other studies have shown that in fact both foraging/exploratory and migratory
movements can potentially lead to dispersal. For instance, (Van Dyck and Baguette, 2005) give
evidence that in some cases, dispersal can be considered a by-product of routine movements. As
for migration leading to dispersal, a well-known example is migratory salmonids: individuals grow
at sea, and when returning to their river of origin to breed, approximately 2 to 10% accidentally
spawn in another river (straying), a mechanism highly contributing to gene flow (Keefer and
Caudill, 2014; Quinn, 1984; Westley et al., 2013). All types of movements are therefore crucial for
animals, at all spatial and temporal scales, in particular because they interact. For instance,
foraging small scale daily movements are crucial to the direct survival of individuals. At a longer
term, migration is determinant for reproduction and therefore for species. Finally, dispersal was
also widely shown to be a key mechanism, driving the spatial and temporal evolution of life-
history traits of species, influencing populations, but also ecosystems. Thus, understanding the
causes and the consequences of movements and the links between types of movements is crucial,
in particular in the era of rapid human induced landscape changes. Indeed, identifying habitats
used for movements, but also individual traits linked to movement can potentially be a prerequisite
for accurate conservation decisions, or at least identification of conservation units at these
different scales.
CONNECTIVITY: A CRUCIAL CHARACTERISTIC OF LANDSCAPES FOR
ALL TYPES OF MOVEMENTS
Because movements are so important for populations, connectivity, i.e. the extent to which a
landscape allows biological fluxes (Baguette and Van Dyck, 2007; Taylor et al., 1993;
Tischendorf and Fahrig, 2000; Uezu et al., 2005), has been a field of high concern in conservation,
but also in fundamental questions on links between movement and landscapes (Fahrig and
Merriam, 1985; Taylor et al., 1993; Ziółkowska et al., 2016), in particular because it is a key
property for the persistence of spatially structured populations (Metzger and Décamps, 1997;
Stevens et al., 2006). There are two facets of landscape connectivity:
17
- Structural connectivity, which consists in shape, size, and relative location of patches/demes, and
is thus a characteristic of the landscape. It therefore depends on physical features such as density
and complexity of between-patch corridors (Beier and Noss, 1998), geographical distance between
patches, and inter-habitat matrix permeability, i.e. the difficulty for organisms to move between
patches (―resistance‖) (Gascon et al., 1999; Martensen et al., 2008; Metzger and Décamps, 1997).
- Functional connectivity, which is the response of individuals to structural connectivity
(movements as well as mechanisms which go with movement such as mortality etc…) as well as
subsequent consequences of movement patterns on populations (observable such as gene flow,
differentiation etc…). Functional connectivity depends not only on the landscape pattern, but also
on the interactions between this pattern and the biological characteristics of the target species, such
as their ability to move in areas of non-habitat (Castellón and Sieving, 2006; Greenberg, 1989).
Therefore, some species/populations can be less affected by fragmentation and have lower
extinction risk than others in the same context of structural connectivity (Sekercioglu, 2002).
Threats to connectivity are therefore a worldwide conundrum for biodiversity (Baguette et al.,
2013), and fragmentation of landscapes is recognized as among the main causes of biodiversity
loss (Kindlmann and Burel, 2008). Indeed, the probability of extinction of a local population is
positively related to its isolation, and negatively related to its size (Baguette et al., 2013; Schnell et
al., 2013): small and isolated populations are more vulnerable to genetic, demographic and
environmental crashes (―extinction vortex‖ (Blomqvist et al., 2010; Fagan and Holmes, 2005)).
Moreover, these smaller and more isolated populations have a reduced rescue potential (Brown
and Kodric-Brown, 1977; Hanski, 1998). Therefore, habitat fragmentation is a direct threat for
species‘ capacity to persist in environments concerned with this phenomenon, thus understanding
how human induced modifications of landscapes affect populations through movements is crucial
for their conservation (Cote et al., 2017; Fischer and Lindenmayer, 2007; Young et al., 1996).
Habitat fragmentation is now affecting ecosystems worldwide (Haddad et al., 2015). The most
famously recognized facet of this problem is probably linked to destruction of tropical primary
forests (Giles and Burgoyne, 2008; Williams, 2006), however, even if more insidious, all
ecosystems are to a certain degree concerned by habitat fragmentation. In rivers, the construction
of dams and weirs is a more recent concern, and its magnitude is also impressive. For instance,
18
only in France, the estimated number of riverine obstacles is of more than 80000 (ROE: référentiel
des obstacles à l'écoulement), and many do not take into consideration connectivity, as they were
built way before this was a concern (Figure 3). However, thanks to numerous studies which have
proven the importance of connectivity, these issues are already taken into account in landscape
management. For example, in 1995, the European Council signed the ―Pan European strategy for
biological and landscape diversity‖. In France, this led to the ―Trame verte et bleue‖ (Alphandéry
et al., 2012), instored in 2007, with the goal of integrating ecosystem functionality in management
projects, notably through preserving ―ecological continuity of landscapes‖ which comprises
―biodiversity reservoirs‖ and ―ecological corridors‖. Information on movements is therefore
constantly needed to better take into account specificities of species requirements in terms of
connectivity.
In rivers, connectivity is very specific: movements are constrained to a one-dimension pathway in
a longitudinal branched network (dendritic network, (Campbell Grant et al., 2007). Moreover, the
river flow is unidirectional, directed downstream (Hänfling and Weetman, 2006; Pollux et al.,
2009). Therefore, the metapopulation dynamics in rivers has been a wild field of research,
showing that upstream populations often serve as sources and downstream populations serve as
sinks (Kawecki and Holt, 2002; Morrissey and de Kerckhove, 2009). Additionally, upper demes
are generally smaller, and have higher slope, thus exhibit more extinction probability, and less
colonization (Gotelli and Taylor, 1999). This is also reinforced by the direction of the flow, which
directs gene flow downstream (Kawecki and Holt, 2002). Connectivity between demes in rivers
has widely been shown to affect movements, and therefore metapopulation dynamics. For
instance, in upstream reaches, populations are generally more differentiated and less genetically
diverse than in lower more connected demes, in which confluence patches are a genetic mix of
several headwater patches, and hence reservoirs for genetic diversity (Carrara et al., 2014; Paz-
Vinas et al., 2015). Thus, headwaters are the sources of diversity in a basin and therefore crucial in
terms of conservation (Campbell Grant et al., 2007). Insuring connectivity in such complex
systems is thus crucial for preserving their particular dynamics, and a prerequisite for this is to
thoroughly understand how they can be affected by natural and/or anthropic perturbations to
connectivity (Blanchet et al., 2010; Dudgeon et al., 2006). Indeed, estimates of connectivity in
rivers have already been shown as highly valuable tools for conservation, particularly because they
19
have successfully enabled to prioritize effort on specific obstacles, or on restoring natural corridors
essential to metapopulation dynamics (Faulks et al., 2011; Prunier et al., 2018; Raeymaekers et al.,
2008).
Figure I3: ROE map of obstacles in river systems of France. Each dot represents an obstacle, color
gradient represents the height (towards red = higher obstacle)
DETECTING/MEASURING MOVEMENTS
Because of their importance for species/individuals/populations, measuring movements has
therefore been the aim of many studies (Holden, 2006; Nathan et al., 2008; Postlethwaite et al.,
2013; Schick et al., 2008), and is quite a challenging task in the wild. Various data types and
methods have been applied to this goal. First, direct methods such as telemetry, PIT tagging, or
mark recapture (Griesser et al., 2014) which are efficient and straightforward, but present
limitations in terms of spatial and temporal scales, number and size of tracked individuals, and
feasibility in general (Cooke et al., 2013; Koenig et al., 1996). Then indirect methods were used,
mainly using molecular approaches. These methods appear very interesting, in particular in this
context for its applicable spatial scale which is coherent with users and managers of these
ecosystems. With the technological advancements in genomics (NGS for instance, but even before
20
with microsatellites) and mass spectrometry for large sample sizes, they are very useful tools in
the study of populations and in particular movements. However, they are more efficient when
inter-population differences are high, and in some cases only utilizable under this condition (Ward
et al., 1994).
Genetic markers: a key approach for large scale studies
Genetics were the first data type to be applied to studying movements indirectly, and still remain
the most used for this purpose. Indeed, population genetics offer a replicable and evolutionary
time-scale data resource which is of very high interest, providing insight into both the ecological
and the evolutionary dynamics of wild populations (Elliott, 1994; Pelletier et al., 2009). Wright
(1931) first started using genetic differentiation as a proxy for the proportion of individuals
moving between populations, through effective sizes (Nem). These methods were shown to have
certain limitations in terms of accuracy (Whitlock and McCauley, 1999). Moreover, it requires
moving individuals to have reproduced in their destination site, and only enables to investigate
gene flow between demes after several generations; therefore real-time movements cannot be
detected. Finally, they require a variety of demo-genetic conditions, such as Hardy-Weinberg
equilibrium, neutrality of markers, absence of linkage disequilibrium etc…But these population-
based methods, along with studies assessing its limitations, opened the door for the development
of many other approaches, and in particular, to individual-based approaches.
The main individual-based approaches to detect between population movements are assignment
methods which were developed and applied to case studies with great success (Bekkevold et al.,
2015; Berry et al., 2004; Bradbury et al., 2015; Karlsson et al., 2011; Manel et al., 2005; Templin
et al., 2011). The overall goal of assignment is to identify immigrants/residents (Pritchard et al.,
2000; Rannala and Mountain, 1997; Wilson and Rannala, 2003), and individuals‘ natal
populations (Cornuet et al., 1999). These methods are highly interesting because they can be used
without the previously required assumptions, and are therefore much more straightforward (Berry
et al., 2004). The first assignment test approach was developed by Paetkau et al. (1995), to study
movements of polar bears, under the null hypothesis that an individual was born in the population
in which it was sampled. This approach consists in calculating the expected frequency of each
individual's genotype in each of the sampled populations and subsequent assignment of each
21
individual to the population where its expected genotype frequency is highest, and is thus the
product of expected genotype frequency at each locus, based on the observed distributions of
alleles. Subsequently, an individual is assigned to the population for which it has the highest
likelihood.
Then, a major challenge is to statistically distinguish real immigrants from Type I error, i.e.
residents which by chance have a genotype more expected in a population different than that in
which they were sampled. For this, different methods exist:
(i) Methods based on likelihood ratios, for instance comparing the likelihood of the individual
genotype within the population where it was sampled versus the highest likelihood value among
all available population samples including the population where the individual was sampled (see
Paetkau et al. (2004) for other likelihood based ratios). Such likelihood ratios are appropriate when
all source populations for immigrants are thought to be sampled.
(ii) Partially Bayesian assignment test (Rannala and Mountain, 1997) that use a Bayesian approach
to estimate population allele frequencies and a frequentist approach to compute the statistical
significance of individual assignments, which were shown to provide slightly higher assignment
accuracy (Cornuet et al., 1999). This test was also extended to include a statistical significance test
based on Monte Carlo simulations, that does not require the assumption that the true population of
origin has been sampled (i.e., an exclusion test).
(iii) A fully Bayesian method (Pritchard et al., 2000) which provides a coherent framework for
incorporating the inherent uncertainty of parameter estimation into the inference procedure. Using
this method, the posterior probability value can be interpreted directly as the probability of origin
of each individual from each population sampled (assuming the true population of origin has been
sampled).
Finally, methods based on discriminant analysis (which defines a model in which genetic variation
is partitioned into a between-group and a within-group component, and yields synthetic variables
which maximize the first while minimizing the second) were also shown very useful for individual
assignment. In particular, Discriminant Analysis on Principal Components (DAPC (Jombart et al.,
2010)), which relies on data transformation using principal component analysis prior to a
22
discriminant analysis. This enables to only perform the discriminant analysis on data that is
perfectly uncorrelated. Another asset of DAPC is that it allows assessing the contributions of
alleles to the structures identified by DAPC, thus making possible to identify regions of the
genome driving genetic divergence among groups for instance.
A strong advantage of these methods compared to estimations of gene flow from differentiation is
that they allow investigating direct time movements: to be detected as a migrant, an individual
does not necessarily have to have reproduced in its new site. This is of high interest because it
enables to detect other types of movements such as migration, or even foraging. This is also highly
informative because even if they do not (yet) participate to gene flow, these movements can be
very important in terms of ecological and populational dynamics. Moreover, these methods are
individual-based, therefore, individual data (phenotype, behavior etc…) as well as environmental
data can also be used for instance when testing mechanisms linked to movement.
Isotopic markers: an alternative approach for measuring movement
The development of assignment methods also enabled a crucial advancement, because they can
also use non-genetic data. Although more recently developed than for genetics, stable isotopes
have been used for a relatively long time for studying movements of animals.
Stable isotopes are indeed indirect markers of high interest for studying animal movements
(Durbec et al., 2010; Hobson and Wassenaar, 2008; Rubenstein and Hobson, 2004). Stable
isotopes are non-radioactive forms (variants) of elements, which differ in nuclear mass. On earth,
about two-thirds of elements occur naturally in multiple stable isotopes. These isotopes are present
in different proportions measured as isotopic differences relative to international standards and
reported as ratios in delta (δ) units as parts per thousand (‰). Variation in these proportions
depends on various geochemical, and biochemical processes, at variable scales, and this is what
can be used to detect movements. The use of this technique to trace the origin or migration of wild
life is based on the fact that stable isotope signatures in animal tissues reflect those of local food
webs or environments either directly, or indirectly through enrichment or other metabolic factors
(Peterson and Fry, 1987): isotopic signatures of food webs or environments can vary spatially and
are passed on to consumers feeding in those food webs/individuals living in those environments
23
(DeNiro and Epstein, 1978). Animals that move between isotopically distinct food webs or
environments can retain information of previous location for periods that depend on the elemental
turnover rate of the tissue considered (Hobson and Clark, 1992; Tieszen, 1991). This can be used
to detect movement provided the spatial and temporal variability in baseline isotopic values are
smaller than those between geographic areas (Rubenstein and Hobson, 2004).
There are two main categories of tissues, on which the turnover rates depend: (i) metabolically
inert tissues such as bones or keratinous tissues (hair, feathers, or nails for instance) which
maintain an isotopic record reflecting the location where the tissue was synthesized (Schell et al.,
1989; Thompson and Furness, 1995), and (ii) metabolically active tissues in which the dietary or
source signature obtained will be a temporal integration ranging from a few days in the case of
liver or blood plasma, to several weeks in the case of muscle or whole blood (―turnover time‖;
(Hobson and Clark, 1992). In addition to a wide variety of usable tissues, different elements and
their different stable isotopes can be used for many different purposes. For the purpose of
determining the origin of animal, they first rely on the distinction between environmental or basal
source signatures. Therefore having a variable baseline across sites is a crucial prerequisite for the
use of isotopes to study movement. Two main types of elements can be used for this purpose
(Rubenstein and Hobson, 2004): (i) heavy isotopes such as Strontium (δ86Sr) or lead
(δ206,207,208Pb), whose abundances tend to be influenced by geological and biogeochemical
processes, and thus vary naturally in soils and bedrock and in some cases as a result of
anthropogenic inputs (Durante et al., 2016; Jiang et al., 2001), and (ii) light isotopes whose
abundance tend to be influenced by both biological and biogeochemical processes (Hayes, 2001;
Kaplan, 1975).
Isotopes have been used successfully in a wide variety of taxa and ecosystems for the purpose of
investigating movement. In particular, δ87Sr, δ13C and δ15N values in both metabolically active
and inert tissues have been widely used to assign individuals to different foraging locations for a
wide variety of taxa. In fish for instance, isotopic ratios of Strontium (δ87Sr) in otoliths which
show annual growth rings that directly reflect isotope ratios of the surrounding water without
fractionation, and can be used to classify individuals to different habitat along their life (e.g.
movements from freshwater to sea, natal stream habitats and the movements of fish from stocked
or locations (Kennedy et al., 2002; Martin et al., 2013; Walther and Thorrold, 2006)). Other
24
examples are numerous, in birds for instance, where isotopes enable to differentiate feeding and
breeding locations (Bearhop et al., 2003; Cherel et al., 2006), but also in mammals (Crawford et
al., 2008; Kelly, 2000; Newsome et al., 2010).
In my PhD, I used stable isotopes of Carbon and Nitrogen in a metabolic active tissue (fin tissues),
which, unlike otoliths, do not require destructive sampling. The turnover rate in these tissues is
around a few weeks (3 to 5, Grey, 2001). Stable carbon (δ13C) and nitrogen (δ15N) isotope values
are influenced directly by biochemical processes during fixation in plants (Chikaraishi et al.,
2004), and particularly during photosynthesis and denitrification processes (Craine et al., 2015). In
terms of fractionation along the trophic chain, δ13C values in animal tissues accurately reflect
those of their diet, i.e. a consumer will have the same signature as the prey (although there can be
enrichment up the trophic chain in some cases, but relatively low 0 to 2‰; (Bocherens and
Drucker, 2003)). Therefore, this isotope ratio can give good information on the type of food
resource of a consumer. On the other hand, δ15N values in animal tissues show considerable
enrichment up the trophic chain (3 to 5‰; (Bocherens and Drucker, 2003)), and can moreover be
affected by water and nutritional stress. They thus reflect the trophic position of an organism.
Therefore, if individuals forage and assimilate food at a particular location and assume isotopic
equilibrium with that diet, then movement patterns between locations with isotopically distinct
basal resources can be inferred (Fry et al., 2003), in consideration of turnover time: more
concretely, Carbon and Nitrogen isotopes can enable to detect an individual‘s displacement if: (i)
the individual stayed long enough (turnover time) in the location it left, (ii) the individual was
sampled recently after its arrival (< turnover time) to its destination location (iii) the isotopic
signatures of the origin and destination locations (as well as the outcome of diet on the
individual‘s tissue signature depending on the trophic conditions) are different. In salmonids, this
has been used successfully by (Rasmussen et al., 2009) for instance, using Carbon isotopes along a
signature gradient in riverscapes: juvenile Atlantic salmon (Salmo salar) exhibited δ13C signatures
different from those of their invertebrate prey at their sampling location, suggesting movements of
~20 km.
More generally, stable isotopes have been used to describe spatial patterns of movement at a
variety of scales (Hobson, 1999; Hobson and Wassenaar, 2008; Rubenstein and Hobson, 2004).
However, the majority of these studies concern large spatial scales and relatively few could be
25
considered at the fine spatial scale (i.e. a few kilometers; Cunjak et al., 2005; Haas et al., 2009;
Litvin and Weinstein, 2004). Moreover, they generally rely on very strong differences between
sites which are also linked to longer distances, therefore explain why these studies often concern
such large spatial scales. For example, δ15N values have been used to study movement between
human imprinted watersheds (high δ15N) and more natural ones (low δ15N: (Cabana and
Rasmussen, 1996), which contrast highly in their isotopic signature. Similarly, Harrod et al. (2005)
also used a salinity gradient to assess site fidelity in eels (Anguilla Anguilla) with δ13C and δ15N.
Within a catchment, isotopic differences may naturally exist between the main river channel and
its tributaries (Douglas et al., 2002), or within a river between reaches (Hershey et al., 1993;
Rasmussen et al., 2009), however, to our knowledge, no studies use stable isotopes of carbon and
nitrogen to study movements within these types of network at fine scales (of a few tens of
kilometers), and in which between site isotopic differences are certainly lower than in the
examples cited previously.
Toward a combined approach
As both genetics and stable isotopes are very valuable tools for detecting individual movements,
using both data types (as well as others potentially) appears as a promising way of improving
detections of movements. Indeed, both data types can capture movements which concern different
time scales, and differences in ―uniqueness‖ of a site / discrimination between sites can vary
between these markers.
Therefore, according to scales considered, combining these tools has the potential for great insight
into movements compared to when only considering one marker. Combining genetics and stable
isotopes has already been done to study movement. For instance in fish, a good example is in
Perrier et al. (2013), where they determine that the progeny of hatchery-born juveniles of Atlantic
salmon released into the wild, which have pure hatchery pedigrees can be determined as river-born
with otoliths isotope analysis, whereas they would be mistaken for hatchery-raised if only genetics
had been considered. In particular, in some cases the spatial discriminant power of a given marker
can be low, notably when the resolution of spatially distinct biological units across fine spatial
scales can be difficult as weak genetic divergence may limit the accuracy of assignment tests
(Larson et al., 2014). Therefore, in these cases, combining datasets can potentially be a highly
26
valuable input across both large and small geographic scales. Some statistical tools developed at
first for genetics such as DAPC (Jombart, 2008; Jombart et al., 2010) can allow using different
types of data to investigate movement, however without easily allowing integration of genetic and
non-genetic data or allowing for the separation of PCs between data types for instance. One of the
latest developments in assignment, which enables to use different data types in combination is
supervised machine learning (Kotsiantis, 2007), which has been raising in terms of interest for
assignment purposes, particularly through an increasing popularity for Random Forests (Chen et
al., 2018; Cutler et al., 2007).
The principle of random forest is very clearly explained by Touw et al. (2013). Briefly, random
forests are ―bagged decision tree models‖ (Breiman, 2001). A decision tree splits data into smaller
data groups based on features of the data until attaining a small enough set that only has data
points under one label. In a decision tree model, these splits are chosen according to a purity
measure (residual sum of squares for regression, Gini index or entropy for classification). Bagging
(or bootstrap aggregating) signifies that a number of decision trees are trained on bootstrapped
training sets. The final predicted value is the average value of all decision trees, or in other words,
the ―majority vote‖. Random forest is an improvement of bootstrap aggregating as it decorrelates
the trees by splitting on a random subset of features: at each split of the tree, the model considers
only a small subset of features rather than all of the features of the model. Therefore, if some
predictors are very strong, trees would be highly correlated, thus random forest enables to go
beyond this limitation. Therefore, predictors can also be ranked by importance based on the
change in classification error affected by the presence or absence of a predictor in a subset, while
considering various combinations of predictor subsets. For the purpose of assigning an individual
to populations, the use of the implemented classification algorithm, where an individual is
assigned to a class (e.g., population) is highly interesting. Random forests are highly promising
because instead of starting with a data model, they use an algorithm to learn the relationship
between the response and its predictors (Hastie et al., 2009).
Random forest is a very advantageous method. Indeed, it exhibits high versatility, handling with
ease either regression or classification, can handle both binary, categorical and numerical features,
and requires no pre-processing such as scaling or transforming the data. As it functions with
subsets of data, random forest is very appropriate for high dimensionality, with impressive
27
prediction speed. Statistically, random forest is a very robust procedure, very efficient at handling
outliers and non-linear data. By averaging trees, variance is also averaged, therefore, random
forest exhibit very low bias. One of most interesting assets of random forest is that it handles
unbalanced data very well, by balancing error according to population size. In empirical datasets,
this is highly interesting because sampling can often be a difficult task, for instance during field
sessions for my PhD, although I aimed at sampling the same number of individuals at each
location, it was in some cases impossible due to sampling conditions or fish density. Another
interesting use of random forests is identification of informative SNPs (Sylvester et al., 2018), or
relative contributions of groups of variables (e.g. genetics vs. isotopes) to explaining the response
variable. However, there are some limitations which must be accounted for: first, random forest
models are relatively intricate, and thus sometimes difficultly interpretable. Another main
drawback is that they can tend to over fit, therefore, choosing the right parameters can be tricky
because of the intricacy. Finally, in the case of very large data sets, the size of the trees can be
quite demanding in terms of computational power.
Overall, random forests appear highly suitable for assignment in the aim of studying movements,
with high sample sizes and high numbers of markers. They have already been used successfully
for assignment and determining population structure, and although it appears intricate in terms of
use, this tool is very promising (Guinand, 2002; Ning and Beiko, 2015; Schrider and Kern, 2018;
Sheehan and Song, 2016; Sylvester et al., 2018). Combining different data types can for instance
enable to answer questions about movements which concern different spatial and time scales than
genetics alone. For instance, when using tissue isotopes, for which turnover time is short (a few
weeks; Busst and Britton, 2018; Franssen et al., 2017), movements will not be accounted for in
the same way than for markers which take several generations to change at the population level
(genetics). Therefore, these markers may also better reflect real time changes in movement during
a recent environmental change. Indeed, there is a recognized time lag between an event causing a
change in genetic structure, and the actual change in structure (Landguth et al., 2010), which can
be misleading when investigating a vicariance object for instance (Cayuela et al., 2018; Prunier et
al., 2014). Assignment methods, using other markers such as isotopes or morphometrics are thus a
promising tool for inferring different types of movements (Chen et al., 2018; Kelly et al., 2005;
28
Rundel et al., 2013), and investigating how these markers and their combinations can improve
assignment tests was one of the aims of this thesis.
THE BROWN TROUT, A FASCINATING SPECIES TO STUDY MOVEMENTS
USING MOLECULAR MARKERS
In river systems, movement is a very important and specific question (Fourcade et al., 2013; Paz-
Vinas et al., 2013). For instance, the European Water Framework Directive (DCE, 2000) defines
the objectives of the European Union to improve ecological status of water bodies. The free
circulation of aquatic organisms and their ability to access to important zones for spawning,
growth, feeding and shelter was reminded to be a key element of a good ecological status. Habitat
connectivity was also law-enforced for natural transport of sediment. The french water legislation,
the LEMA (loi sur l‘eau et les milieux aquatiques Art. L214-17, 2006) even protects some rivers
from new obstacle constructions and some existing ones must be adapted to fish and/or sediment
circulation. As connectivity is a great concern in these environments, and laws are being written to
protect it, it appears essential to deepen knowledge on this mechanism in order to optimize its
enforcement. Fish have in fact been the main trigger for enforcing these new laws on riverine
connectivity. Indeed, the most obvious species concerned are diadromous migratory fish. Salmon
for instance, need to circulate between rivers and sea to complete their life cycle. Therefore,
fragmentation has a direct effect on these species. However, concern was also raised for resident
(or partially migrating) organisms, as the effects of fragmentation are now better understood.
Among the numerous species inhabiting rivers, the brown trout (Salmo trutta) is one of the most
widespread freshwater fish species in Eurasia, and it has been widely introduced in both the
southern and northern hemispheres (Lobón-Cervía and Sanz, 2018; Baglinière and Maisse, 1999;
Rahel, 2007). As all Salmonidae, it is a scientifically interesting species because of its diversity in
terms of ecology, life history strategies and habitat use (Rahel and Nibbelink, 1999; Klemetsen et
al., 2003). The brown trout was shown to have evolved mainly because of the multiple Pleistocene
glaciations (isolation by glaciers and subsequent dispersal and secondary contact between lineages
(Sanz et al., 2000)): during the last 1.7 million years, there were 17 episodes of glaciation-
deglaciation, with two periods of freshwater fish isolation in each glacial cycle. This gave many
opportunities for geographical isolation and divergent evolution (Hamilton et al., 1989), because
29
glaciation cycles affected climate, habitats and the potential for dispersal, translating into marked
patterns of genetic diversity and structure. Seven main mitochondrial DNA lineages were shaped
(although more may be to be discovered): Atlantic (AT), Mediterranean (ME), Danubian (DA),
Adriatic (AD), Marmoratus (MA), Duero (DU) in Europe, Tigris (TI) in Turkey (Bernatchez,
2001; Susnik et al., 2005; Vera et al., 2010), and Dades in Morocco (Snoj et al., 2011) and have
remained relatively allopatric since then for both biological reasons and physical isolation (Figure
4).
However, some sub-lineages have been identified throughout Europe, making the situation far
more complex. For example, Northern Africa harbors one AT matriarchal lineages (Tougard et al.,
2018) that diverged significantly from the main AT lineage identified by Bernatchez (2001).
Among these lineages, there is also high divergence. For instance among the Atlantic group,
several sub-lineages were discovered (Cortey and Garcia-Marin, 2002; Krieg and Guyomard,
1985; Vera et al., 2015). Similarly, the Western Mediterranean regions host divergent sub-lineages
of the AD and ME evolutionary lineage between the Iberian Peninsula (Cortey et al., 2004), and
Italy (Splendiani et al., 2016). Finally, island sub-lineages have also been described as divergent
from the continental populations, for instance the Corsican populations (Berrebi et al., 2019).
Figure I4: phylogeny and range of the 5 main brown trout lineages
30
At more local scales, the brown trout exhibits very high levels of variability in terms of genetic
diversity and population structure, but also in terms of ecology (Swatdipong et al., 2010; Vøllestad
et al., 2012; Berrebi, 2015). This species thus appears as the perfect model to study dispersal in a
human affected landscape. Indeed, inhabiting higher river stretches, among those most concerned
by fragmentation, this species is often the one for which connectivity efforts must be made, in
particular because of its mobility (Gouraud et al. 2008) and its high association to human interests
(Butler et al., 2009; Ferguson, 1989; Mills, 1989).
STOCKING PRACTICES IN EUROPE, AND EFFECT ON MOVEMENT
INFERENCE
A strong bias to the study of movements in natural systems is human-induced displacements or
introductions in these systems. Because the brown trout is strongly associated to human interests,
wild brown trout populations are widely managed and this species has been domesticated for more
than a century, and spread worldwide (Laikre et al., 1999). Hatchery strains have been largely used
to sustain wild populations worldwide (Elliott, 1994; Berrebi et al., 2000; Bohling, 2016), mainly
from the Atlantic strain (majority of Danish origin; Laikre et al., 1999), even in rivers naturally
inhabited by Mediterranean strains (Krieg and Guyomard, 1985). Therefore, these massive
introductions in river systems make the study of movement, especially using genetics, very
difficult. Thus, identifying introduced vs. natural strains is an indispensable step, and measuring to
what extent the natural spatial patterns are affected by this practice is a prerequisite for
understanding natural populations processes.
This is a very difficult task, indeed, although stocking started with mainly Atlantic trout, (Hansen,
2002; Araki and Schmid, 2010; Christie et al., 2012; Leitwein et al., 2016), its negative effects
(Largiader and Scholl, 1996) led managers to change stocking practices (Arlinghaus and Mehner,
2005; Laikre et al., 1999) either by stopping it, or by using strains closer to the ones naturally
found in the rivers stocked (Caudron et al., 2006). There has thus been such a long period of
stocking, with many strains from different origins that they are now part of the natural dynamics of
populations (Saint-Pé et al., 2018). Captive-breeding and the selection pressures associated to it
generally make these strains very different from wild ones (Blanchet et al., 2008; Christie et al.,
2016; Frankham et al., 1986), thus enabling to distinguish wild from non-wild individuals.
31
However, tracking the stocking histories is extremely complex, and some populations used to
stock are no longer available for sample, therefore, information can be lacking in the case of
multiple hatchery stocks.
MOVEMENT PATTERNS IN BROWN TROUT POPULATIONS
In a river system, all fish are not mobile, and movements are highly variable among individuals,
populations and depend on the species considered (Radinger and Wolter, 2014; Skalski and
Gilliam, 2000). Moreover in river systems, the characteristics of the network as well as
hydrological features (slope, flow, speed of flow, obstacles etc…) shape very particular patterns of
movement (Carrara et al., 2014; Hitt and Angermeier, 2008; Labonne et al., 2008; Paz-Vinas et al.,
2015). For instance, upstream reaches, which are more isolated, and smaller tend to have less
immigrants than more connected ones, with strong genetic implications such as strong isolation by
distance (IBD, i.e. pairwise genetic differentiation increase with riparian distance between
populations) and downstream increase in genetic diversity (Paz-Vinas and Blanchet, 2015).
The brown trout exhibits very complex movement patterns which strongly depend on spatial, and
time contexts (Baglinière, 1999; Elliott, 1994; Jonsson et al., 1991). Aquatic ecosystems inhabited
by brown trout present a continuous gradient of physical conditions (river continuum concept;
Vannote et al., 1980), brown trout can thus exhibit a continuum in time and space of life history
tactics to optimize individual fitness and population persistence (Cucherousset et al., 2005). Three
main ecological forms of brown trout have however been described (Hindar et al., 1991;
Klemetsen et al., 2003), and although they are genetically undistinguishable, the strategies were
shown to be heritable (Ferguson, 2007): resident river trout, sea trout (which migrate to sea for
growth), and lake trout (which migrate to lakes for growth), all forms spawning in rivers. This
thesis only concerns the first of these forms, as the rivers studied do not have access to the sea or
to lakes. Cucherousset et al. (2005) showed that depending on a combination of individual (age,
growth) and environmental factors, movement patterns are highly variable even at a very small
spatial scale, and can take all possible forms. Investigating movement patterns in this species at a
broader scale is thus a challenge and disentangling the different types of movement is very
difficult.
32
Life cycle of the brown trout
In the brown trout, sexual maturity is attained around 2 years old (Maisse and Bagliniere, 1990).
Reproduction generally occurs from September to January (Gouraud 1999) after a spawning
migration, generally upstream towards tributaries by adults (Elliott, 1994; Höjesjö et al., 2007;
Whoriskey, 2005). The number of eggs depends on the size of the female, for instance a 500g will
ley around 800 eggs. Then, depending on conditions, between 4 and 80% of these eggs will hatch,
generally around February, after a duration depending on water temperature: 60 days at 7.8°C, and
97 days at 4.7°C. Then the alevins live in the gravel and feed off the remaining yolk for two to
four weeks also depending on temperature. Once the yolk has been consumed, the alevins become
fry: they emerge from the gravel and start feeding on small invertebrates in slow flowing shallow
water (1 – 40cm). After their first year (―parr‖ stage), and as they grow, they chose deeper and
faster water, gradually dispersing downstream (Bagliniere et al., 1989; Northcote, 1992; Solomon
and Templeton, 1976; Olsen and Vollestad, 2005; Schager et al., 2007), although rare upstream
movements of juvenile trout have been observed on very small distances (Heland, 1980; Lucas et
al., 2001). Mortality rates at these vulnerable young stages are high. In terms of population
dynamics, this cycle therefore generally translates in tributaries acting as sources (reproduction of
adults, and departure for juveniles), while mainstreams act as nurseries and feeding/growth zones
(Figure 3), although this simplified model is very context dependent (Acolas et al., 2008;
Bagliniere et al., 1989; Bembo et al., 1993; Cuinat and Heland, 1979; Dieterman and Hoxmeier,
2011; Gowan and Fausch, 1996; Klemetsen et al., 2003; Maisse and Bagliniere, 1990; Ovidio et
al., 1998).
33
Figure I5: Simplified life cycle of resident Brown trout. Adults living in lower stretches of the mainstream
spawn in tributaries after an upstream migration. After the hatch, the juveniles leave the nursery areas and
start dispersing downstream.
Movement determinants in the brown trout
Movement in this species is highly variable, and appears very context dependent as some studies
support that movements are very limited (Bagliniere et al., 1989; Harcup et al., 1984; Knouft and
Spotila, 2002; Rodríguez, 2002) whereas others show high mobility (Gowan and Fausch, 1996;
Höjesjö et al., 2007). The types of movements can also exhibit high variability, for instance
foraging movements generally concern small areas (~15 m2; Bachman, 1984) although they can
also take place on very long distances (>30 Km; Clapp et al., 1990), while dispersal of young of
the year can extend on several kilometers although it generally occurs on a few hundred meters
(Aparicio et al., 2018; Knouft and Spotila, 2002; Vøllestad et al., 2012). Spawning migrations are
also variable, and depend on the context, but generally concern a few hundred meters to a few
kilometers (Burrell et al., 2000). Finally, some extreme movement events (more than 120 Km) of
which the cause was not identified have been reported (Linløkken, 1993). Determinants of
movement in the brown trout are complex, and strongly interact with the developmental stage, as
the different spatial uses in a river basin depend on these stages (Roussel and Bardonnet, 2002;
34
Acolas et al., 2008; Cucherousset et al., 2005; Forseth et al., 1999; Greenberg and Giller, 2001;
Heggenes and Traaen, 1988; Olsen and Vollestad, 2005; Vollestad et al., 2002) in which
tributaries act as nurseries for juveniles, while mainstreams mostly host larger adults (Bembo et
al., 1993; Kelly-Quinn and Bracken, 1989; Maisse and Bagliniere, 1990).
From the first stage following emergence (fry), brown trout already become territorial and a
hierarchy is put in place, through competition for habitat: fry emerging later, or not able to rapidly
find suitable habitat were shown to drift farther downstream for instance (Cuinat, 1971; Heland,
1980). Later in the life cycle, dominance relationships are also shaped by aggressive behavior and
individual size, as individuals defend a well-defined home range forcing subordinates fish to move
(Elliott, 1994; Höjesjö et al., 2007). Larger and more dominant fish also tend to have larger home
ranges than smaller and subordinate ones, and feed during the most beneficial times of the day
(Alanärä et al., 2001). These relations therefore occur among the different size/age classes in the
population (Skalski and Gilliam, 2000) and also between males and females (Hutchings and
Gerber, 2002; Pirhonen and Forsman, 1998). Metabolism is also very determinant at all stages, for
instance it was shown individuals with a faster metabolism, and therefore faster growing, emigrate
earlier downstream at the fry and parr stages (Cucherousset et al., 2005; Forseth et al., 1999).
Moreover, in adults large and faster growing fish move more, which is congruent with the fact
they can more easily afford the costs of movement (Bonte et al., 2012). However, metabolism is
closely linked to growth and thus size, and therefore, Heland (1980) found that fry with lower
growth rate tended to move more, as smaller individuals. Once again, movement patterns in this
species are highly variable and context-dependent, and different determinants of movements can
strongly interact.
Environmental conditions have been shown to affect all components of movements in trout. In
foraging movements for instance, lateral within ―home range‖ movements depend mostly on
nychtemeral patterns (Roussel and Bardonnet, 2002; Heland 1980). Environmental conditions
appear to be relatively context dependent, for instance, habitat characteristics such as depth, cover
and water velocity were shown to affect movement, but differently according to the river
considered (Aparicio et al., 2018). Water flow is also a crucial determinant for the dispersal of
juveniles (Gouraud et al., 2008; Holmes et al., 2014), particularly because the downstream
35
dispersal of fry is in large part due to their low swimming ability, and is therefore passive (Crisp
and Hurley, 1991; Cuinat and Heland, 1979; Heggenes and Traaen, 1988). Moreover, the type of
stream is also determinant in these movements, as they generally consist in movements from
tributaries to the mainstream (Bembo et al., 1993; Dieterman and Hoxmeier, 2011), and
additionally, they were also shown to be determined by higher habitat diversity and food
availability in these downstream destination reaches (Aparicio et al., 2018; Northcote, 1992).
Indeed, as fish grow larger, preferences in cover, water depth and velocity and substrate size may
change, leading to niche shifts (Heggenes et al., 1999; Ovidio et al., 1998). Finally, spawning
migrations were also show to be driven by environmental factors (Cucherousset et al. 2005). These
migrations are the most important movement in this species (García-Vega et al., 2017). They are
determined by multiple factors such as solar radiation, flow and water: in particular, low
temperatures and low and stable flows were shown to trigger these migrations (Slavik and Bartos,
2004). Finally, food resource also plays an important role in these migrations as migrating towards
productive feeding habitats was shown to increase fitness, notably in females because it allows for
higher fecundity (Northcote 1992)
Populational parameters are also determinants of movements; indeed, both population density and
kinship between individuals were shown to act on dispersal in particular. First, population density
was shown to be an important trigger of movement. In particular, during the fry and parr stages,
dispersal was shown to be density-dependent in many studies as individuals flee over populated
locations in which competition for space and resource is too high (Bagliniere et al., 1989; Cuinat
and Heland, 1979; Einum et al., 2006; Baglinière and Maisse 2002). Second, kin interactions and
inbreeding were also shown to modulate movements (Sanz et al. 2011), for instance, in poor
environmental conditions, dispersal of related fish was shown to be low thus resulting in high
relatedness among neighboring juveniles (Hansen et al., 1997; Vera et al., 2010). This is
unexpected as dispersal was generally shown to be a way of avoiding kin-interaction via kin-
avoidance (Kasuya, 2000; Perrin and Mazalov, 1999). However, kin-biased distribution can also
present many benefits such as increasing the chances that at least one or a few relatives spawn, and
has been observed in several studies on salmonids (Carlsson et al., 2004).
36
The importance of connectivity in trout movements
These complex movement patterns, necessary to the species‘ life cycle, are however affected by
habitat fragmentation. The impacts of dams/weirs were widely studied, and shown to affect
different types of movements all across the life cycle. First, and most obviously, all upstream
movements can be impeded or at least made very difficult when an obstacle is present. Therefore,
this affects mainly adult migration (Gosset et al., 2006; Ovidio et al., 1998). Moreover, subsequent
flow regulation by dams also affect migration patterns by providing non-natural high flows and
colder water during summer (García-Vega et al., 2017). However, studies have shown that
downstream movements are also affected by dams, in particular downstream dispersal of
juveniles, as dams were shown to increase mortality at this development stage (Larinier and
Travade, 2002). Therefore, dam users are increasingly taking these issues into account, equipping
them with systems enabling upstream passage, but also downstream (Larinier and Travade, 2002;
Schilt, 2007), which have already shown to be successful (Tomanova et al., 2018).
37
AIMS OF THE PHD
The overall aim of this thesis is to develop new methodological approaches to study movement at
the scale of a river basin, and to bring new perspectives on movement patterns and their
underlying causes. This thesis therefore has a double aim: a methodological aim which consists in
developing novel tools to study movement, and a more biological aim to bring new insight in the
biology of movement.
The first aim was answered with both the development of a universal genetic tool, and then the use
of recent statistical developments to assess how combining different markers to study movement
can improve accuracy and power. Genetics and isotopes notably are very valuable tools to study
movements; however, they are for now mainly used separately, and on spatial scales which require
a high differentiation between sample locations to make movement inference possible. Therefore,
using new methods of machine learning to combine them is a very promising approach, and at the
scales of the rivers studied in this PhD, these novel methods appear as an optimal way of
investigating movements. This is notably of high interest for conservation, as in general, managers
focus on rather local spatial scales. Indeed, users and managers of river systems, such as land
owners, dams and weirs operators, and actors of fish/river management (AFB, DDT, Dreal,
Angling departments and associations, and environmental agencies) are generally interested in
what is happening at the scale they are managing. This methodological aim is developed in
Chapters 1 and 2:
- In chapter 1, I developed a tool useful for detecting genetic structure in the brown trout, as well
as hybridization with captive-bred strains. Although many tools exist for population genetics in
this species, in particular microsatellites and mitochondrial markers, they appear costly and time
consuming in the era of new generation sequencing, and especially lack replicability between
studies. We therefore developed Single Nucleotide Polymorphism markers (SNPs), by first
identifying a set of 12204 RADs markers mapped on the brown trout linkage map and then used
this novel resource to develop a cost-effective array of 192 SNPs evenly spread on the brown trout
linkage map.
- In the second chapter, we use recent statistical developments to assess how combining different
markers to study movement can improve accuracy and power. More precisely, we assessed the
38
contribution of combining genetics, carbon and nitrogen stable isotopes, and morphometrics in
assigning individuals to their population of origin in two independent river catchments. For this,
we used a machine learning framework, based on random forest classification to determine
individual probabilities of originating from each sampling site, using every combination of our
data (genetics, isotopes, morphometrics, genetics + isotopes, genetics + morphometrics, isotopes +
morphometrics, and genetics + isotopes + morphometrics), and implemented a permutation
procedure which enabled to obtain a threshold above which assignment can be considered robust.
This chapter also highlights which types of data and combinations can be of use at such spatial
scales, at which differentiation between sampling locations can be relatively low with one data
type, but better resolution can be obtained by combinations.
The second goal is biological, and consists in improving knowledge on the brown trout,
particularly in the context of human-induced perturbations: the final aim being to identify factors
underlying movement, we were first confronted to the effects of stocking wild populations with
captive-bred strains. Indeed, this practice was shown to modify spatial patterns of genetic diversity
and differentiation, which are the base of movement detection, therefore, stocking can strongly
interfere with movement inference. Therefore, before studying movements and their determinants,
we first described the genetic effects of captive breeding. Movements have been widely studied at
extremely small scales for instance with direct measurements (in particular with many studies on
daily movements using PIT tags, downstream dispersal of fry after hatching from tributaries to
mainstream), or on the contrary at very wide scales using genetics (phylogeny of lineages for
instance) or isotopes but for very contrasted environments and long distances (anadromous-
freshwater migration for instance), but much less at ―river basin scales‖, which, as discussed
above, is the scale of interest for many organizations managing rivers, in particular those
managing brown trout populations. This part therefore aims at better understanding how factors
influence movements, which, in term, aims at better quantifying what is needed for improving
connectivity in this species, and to prioritize conservation efforts. Indeed, for preserving facets of
trout life history linked to movements, such as ―ecological continuity‖, information on movements
at the scale of interest is constantly needed. These aims are developed in chapters 3, 4 and 5:
- In chapter 3, we first investigated how stocking affects spatial patterns of genetic diversity and
differentiation. Understanding how stocking rivers with captive-bred fish can affect these natural
39
patterns at the scale of a river basin is therefore a crucial question. Using the previously developed
192 SNP panel (Chapter 1), we aimed at quantifying and described the spatial distribution of
captive-bred ancestry in four independent watersheds having been stocked using Bayesian
clustering methods, and tested the effect of captive-bred ancestry on the distribution of genetic
diversity and differentiation.
- In Chapter 4, we investigated how patterns of admixture between wild and captive-bred
individuals affect patterns of dispersal in a brown trout population from a small watershed that was
stocked until 1999. We inferred first generation migrants to identify dispersal events and test
whether genetic admixture (in addition we tested the effect of sex and body length) affected
dispersal probability, distance of dispersal and direction of the dispersal event.
- Finally, in chapter 5, we investigated the individual, environmental, and populational
determinants of movements in three riverscapes. Understanding patterns of connectivity/dispersal
is of crucial importance in evolution, ecology and conservation biology, and assessing how
different components (individual, environments, populations) interact in shaping dispersal patterns
is thus of high interest for both fundamental and applied issues. Using assignment methods based
on random forests, with genetics (SNPs developed in chapter 1 and stable isotopes of Carbon and
Nitrogen) as developed in chapter 2, we investigated the effect of individual phenotypes,
environmental characteristics, and populational indicators in shaping movement patterns in three
independent river basins. More precisely, we tested whether individual characteristics are linked to
dispersal propensity and movement characteristics. Then, we investigated which populations are
more likely to receive immigrants or contrary to be emigrated from, linking these percentages to
environmental or/and populational drivers. Finally, using a pairwise approach, we tested whether
individual exchanges between two sites are linked to connectivity and to environmental similarity.
40
Chapter I. Development of a large SNPs resource and a low-
density SNP array for brown trout (Salmo trutta) population
genetics
Keoni Saint-Pé
1#, Maeva Leitwein
2,3#, Laurence Tissot
5, Nicolas Poulet
6, Bruno
Guinand 3, Patrick Berrebi
3,4, Geoffrey Marselli
1, Jean-Marc Lascaux
7, Pierre-
Alexandre Gagnaire 3*
, Simon Blanchet 1,8*
#co-first authors
*co-last authors
1 Centre National de la Recherche Scientifique (CNRS), Université Paul Sabatier (UPS); Station d‘Ecologie
Théorique et Expérimentale, UMR 5321, F-09200 Moulis, France
2 Département de Biologie, Institut de Biologie Intégrative et des Systèmes (IBIS), Université Laval,
Québec, QC, Canada
3 ISEM, CNRS, Univ. Montpellier, IRD, EPHE, Montpellier, France
4 Present address: Genome - Research & Diagnostic, 697 avenue de Lunel, 34400 Saint-Just, France
5
EDF R&D LNHE - Laboratoire National d‘Hydraulique et Environnement, 6 quai Watier, 78401 Chatou
Cedex, France
6 Pôle écohydraulique AFB-IMT, allée du Pr Camille Soula, 31400 Toulouse France
7 ECOGEA, 352 Avenue Roger Tissandié, 31600 Muret, France
8 Université de Toulouse, UPS, UMR 5174 (EDB), 118 route de Narbonne, F-31062 Toulouse cedex 4,
France
Key words: conservation, management, population genomics, microarray, admixture.
Manuscript published in BMC Genomics, volume 20, Article number: 582 (2019)
DOI : 10.1186/s12864-019-5958-9
41
Abstract
Background The brown trout (Salmo trutta) is an economically and ecologically important species
for which population genetic monitoring is frequently performed. The most commonly used
genetic markers for this species are microsatellites and mitochondrial markers that lack
replicability among laboratories, and a large genome coverage. An alternative that may be
particularly efficient and universal is the development of small to large panels of Single
Nucleotide Polymorphism markers (SNPs). Here, we used Restriction site Associated DNA
sequences (RADs) markers to identify a set of 12,204 informative SNPs positioned on the brown
trout linkage map and suitable for population genetics studies. Then, we used this novel resource
to develop a cost-effective array of 192 SNPs (96 x 2) evenly spread on this map. This array was
tested for genotyping success in five independent rivers occupied by two main brown trout
evolutionary lineages (Atlantic -AT- and Mediterranean -ME-) on a total of 1862 individuals.
Moreover, inference of admixture rate with domestic strains and population differentiation were
assessed for a small river system (the Taurion River, 190 individuals) and results were compared
to a panel of 13 microsatellites.
Results A high genotyping success was observed for all rivers (<1 % of non-genotyped loci per
individual), although some initially used SNP failed to be amplified, probably because of
mutations in primers, and were replaced. These SNPs permitted to identify patterns of isolation-
by-distance for some rivers. Finally, we found that microsatellite and SNP markers yielded very
similar patterns for population differentiation and admixture assessments, with SNPs having better
ability to detect introgression and differentiation.
Conclusions The novel resources provided here opens new perspectives for universality and
genome-wide studies in brown trout populations.
42
Background
The brown trout (Salmo trutta) is one of the most widespread freshwater fish species in Eurasia,
and it has been widely introduced in both the southern and northern hemispheres (Lobón-Cervía
and Sanz, 2018). As part of the Salmonidae family, it is a scientifically interesting species because
of its diversity in terms of ecology, life history strategies and habitat use (Klemetsen et al., 2003;
Rahel and Nibbelink, 1999). Thanks to its wide ecological variability and excellent ability to
spread and colonize new watersheds, the species is found both in fresh and salt waters over most
of its range. The brown trout is also an economically major species in terms of farming, net fishing
(for the sea-run form), and expenditure in recreational angling (Butler et al., 2009; Mills, 1989),
partly explaining its worldwide intentional introduction (Rahel, 2007). Because this species is
strongly associated to human interests, wild brown trout populations are widely managed, either to
sustain attractive leisure activities such as recreational angling or to conserve declining and/or
emblematic populations. Moreover, the brown trout has been domesticated since the 19th
century
(Laikre et al., 1999), and hatchery strains have been largely used to sustain wild populations
worldwide (Elliott, 1994; Berrebi et al., 2000; Bohling, 2016). Instead of positive expected effects
of these stocking activities, most have proven to have negative long term effects on wild brown
trout populations in part because of the reduced fitness brought by hatchery fish in wild
populations, and the loss of local genetic heritage caused by the replacement of local wild
populations with genetically homogeneous hatchery strains (Hansen, 2002; Araki and Schmid,
2010; Christie et al., 2012; Leitwein et al., 2016).
The brown trout presents high levels of phenotypic and genetic polymorphism, with seven main
mitochondrial (mtDNA) lineages with various geographical extents being generally recognized.
These consist of (i) four sub-continental lineages: the Atlantic (AT), Mediterranean (ME),
Danubian (DA) and Adriatic (AD) lineages, (ii) two regional lineages: Marmorata (MA) in the
north of the Adriatic Sea and North African (NA) in Morocco, Algeria and Sicily (Bernatchez,
2001; Susnik et al., 2005; Tougard et al., 2018) and (iii) two local lineages limited to one river
basin: the Duero (DU) in Europe, and Tigris (TI) in Turkey (Bardakci et al., 2006; Suárez et al.,
2001; Vera et al., 2015, 2010). Within these extent lineages that cover the whole range of S. trutta,
high levels of genetic and phenotypic polymorphism are also observed at more local spatial scales
within lineages (e.g. population scale; Berrebi, 2015; Swatdipong et al., 2010; Vøllestad et al.,
43
2012). However, diversity patterns in brown trout have also been locally influenced by stocking
practices that mostly relied on European hatchery strains of AT origin (Cortey et al., 2004) to
supplement local populations, with the exception of a few local strains stemming from local
populations (Bohling et al., 2016; Caudron et al., 2006; Cortey and Garcia-Marin, 2002).
Genetic tools appeared as a key approach for scientists and local managers to optimize
conservation efforts (Altukhov et al., 2000; Sato and Harada, 2008) because they provide insight
into both the ecological and the evolutionary dynamics of wild populations (Elliott, 1994; Pelletier
et al., 2009). For instance, assignment tests, fine-scale population structure, kinship analyses and
genome-wide surveys (Araki et al., 2007; Stinchcombe and Hoekstra, 2008; Bourret et al., 2013)
enable to monitor populations effectively, and have high potential applications for conservation
and management in salmonids, including the brown trout (Caudron et al., 2006; Hansen, 2002;
Waples and Hendry, 2008).
Molecular studies on trout populations first used allozymes, mitochondrial markers, and then
microsatellite loci (Apostolidis et al., 1997; Carlsson et al., 1999; Laikre et al., 1999). These
markers (notably microsatellites) are useful and adequate to answer many biological questions, but
their genome coverage is generally weak, and replicability and universality are relatively low since
each research group generally uses its own panel of markers. Single nucleotide polymorphisms
(SNP) markers have been shown to potentially reduce these limitations (Davey et al., 2011; Etter
et al., 2011; Helyar et al., 2011). They allow to uncover a relatively high number of annotated and
mapped markers with low scoring-error rates (Brumfield et al., 2003; Morin et al., 2004). Also,
SNPs markers can easily be chosen to represent both neutral genomic regions and regions under
selection, at a genome-wide scale and across large samples (Morin and Mccarthy, 2007). Despite
being biallelic markers, SNPs can be highly informative for most analyses used in population
genetics, as far as the number of loci is sufficiently high and evenly spread across the genome
(>50; Bradbury et al., 2015; Freamo et al., 2011; Kaiser et al., 2017; Karlsson et al., 2011;
Paschou et al., 2007). Genome coverage is an important aspect in the choice of markers for
population genetics: first, markers evenly spread across the genome are less likely to exhibit
linkage disequilibrium (Scheet and Stephens, 2008), and second, it was shown that many
population events, such as introgressive hybridization can only concern certain genomic blocks, a
full coverage thus enables to capture these events (Lamaze et al., 2012; Leitwein et al., 2018).
SNP arrays are commonly used for conservation purposes in salmonids (Dominik et al., 2010;
44
Kochzius et al., 2008; Lien et al., 2011), although rarely in brown trout, which is probably because
only a handful of SNP markers were available for this species (Pustovrh et al., 2012; Drywa et al.,
2013; Sušnik Bajec et al., 2015). However, higher density resources were more recently
developed. In particular, Linløkken (Linløkken et al., 2017) developed 3781 SNPs to analyse
genetic differences between wild and hatchery brown trout in a tributary of Lake Savalen in
central Norway. Moreover, a new lineage-specific high density linkage map for S. trutta
comprising ancestry informative SNPs for both Atlantic (AT) and Mediterranean (ME)
evolutionary lineages from Western Europe was also developed (Leitwein et al., 2017, 2016). This
latter resource provides a novel baseline for the development of mapped SNPs that may be of
prime interest for studies on brown trout population genetics across a wide spatial range.
The aim of this study was to develop a genome-wide, mapped and universal set of SNPs for the
brown trout for both Atlantic and Mediterranean lineages (AT and ME lineages; sensu Bernatchez,
2001), which would be a useful and affordable tool for both scientists and environmental
managers. By taking advantage of the new genomic resource available for brown trout (Leitwein
et al., 2018, 2017, 2016), we developed a panel of 12,204 RADs containing 1 or 2 polymorphic
SNPs evenly spread across the genome structured in 40 chromosomes, and ancestry informative
for at least two of the main brown trout lineages, the Atlantic and the Mediterranean lineages.
Among these, a sub-panel of 192 SNPs describing the whole genome was included in a low-
density SNP array. The validity of this low-density SNP array was tested by quantifying
genotyping success and number of polymorphic loci in five independent populations from the two
lineages at a large spatial scale. Finally, the power of this array was compared to a panel of 13
microsatellites for answering classical population genetics questions: admixture with stocked
domestic individuals and genetic differentiation among wild populations. All resources are made
freely available for future users (Appendices S1 and S2).
Methods
Development of a genome-wide reference SNP panel
The panel of SNPs markers identified is this study was filtered from the variants described in
Leitwein (Leitwein et al., 2018) using double-digest RAD sequencing. Restriction enzymes EcoRI-
45
HF and MspI were used to digest individual genomic DNA and create the dd-RAD library, which
was submitted to size selection in order to retain fragments of 200 to 700 bp using CleanPCR
beads. The library was then amplified by PCR and sequenced with Illumina HiSeq2500, producing
125-bp paired-end reads. The initial set contained 75,000 SNPs discovered from 82 wild
Mediterranean S. trutta from tributaries of the Orb River catchment in southern France, and 102
captive-bred individuals from farms formerly used for stocking in this region (41 and 61 hatchery
fish from the ME and AT lineages respectively; see Leitwein (Leitwein et al., 2018) for details).
These SNPs were anchored to the-high density S. trutta linkage map using an intermediate step of
physical mapping to the Atlantic salmon reference genome: using their relative positions on the
Atlantic salmon reference genome, it was possible to determine the relative mapping positions of a
large number of additional RAD loci that were not present on the brown trout linkage map
(Leitwein et al., 2017).
We applied a series of filters which allowed for selecting a panel of SNP markers that were (i)
likely to be highly polymorphic, (ii) mapped on the linkage map, and (iii) present in the two brown
trout lineages. To do so, we removed from the initial database all SNPs with a minimum allele
frequency (MAF) of 5% or less using vcftools (Danecek et al., 2011) based on all individuals (AT
+ ME). Then, in order to have clean sequences, straightforward to be genotyped, RAD sequences
with more than two SNPs, and SNP positions falling in the first or last 30 bp of the RAD
sequences were also removed, as well as sequences with undetermined (N) nucleotides. Finally,
we kept only RAD sequences for which mapping positions on the S. trutta linkage map were
determined (see above).
Development of the low-density SNP array
We used this large SNPs resource to develop a low-density SNP array containing 192 SNPs. The
goal was to propose a cost-effective tool which holds on two 96-wells genotyping plates for
analyzing a large number of individuals. Moreover, SNPs were selected to be informative for
population genetics analyses of brown trout populations from at least the AT and ME lineages
(sensu Bernatchez (Bernatchez, 2001)) from Western Europe. The array was genotyped using the
KASPAR technology ® (Smith and Maughan, 2015) that allows a rapid and cost-effective
genotyping service for such a number of markers. We hence applied further filters to fulfill
KASPAR genotyping constraints. In particular, RADs with more than one SNP between the
46
primer designing zones (50bp at the two extremities) were excluded. To ensure a good
representativeness of the genome, we selected SNPs evenly spread and spaced by at least 3.5 cM
across the 1453 cM estimated length of the S. trutta linkage map (Leitwein et al., 2017). This
resulted in 245 SNPs (average 1453/245=5.9 cM) among which, in order to retain 182 SNPs, we
randomly removed 63 SNPs using the sample() R function. Then, we added five ancestry
informative SNPs developed by the Institut National de Recherche en Agronomie and the
University of Savoie (UMR-0042 CARRTEL, France) that were used in previous studies to
distinguish individuals from the AT and ME lineages (OMM1164, OMM1105, OMM1154,
Str541INRA, Str591INRA; Harrang et al., 2015; Caudron et al., 2012, 2006; Estoup et al.,
2000).Finally, we added five mitochondrial SNPs previously used to differentiate among the five
main brown trout lineages (mitoDA10Proline, mitoDA10ProlineB, mitoCytoB, mitoATPaseIVA,
mitoATPaseIVB; (Bernatchez et al., 1992; Giuffra et al., 1994); Appendix S1). These numbers of
five ancestry informative nuclear or mtDNA markers were chosen to represent approx. 5% of the
markers present on the SNP array. This resulted in a low-density array of 192 SNPs markers.
Information and sequences are available on Figshare, DOI: 10.6084/m9.figshare.8174708
Genotyping success of the low-density SNP array
This low-density SNP array was first evaluated for genotyping success using individuals from
five independent French river basins: two from the Pyrénées mountains (the Ône and the Aude
Rivers), two from the Alps mountains (the Roya and the Doron de Bozel Rivers) and one from the
Massif Central mountains (the Seuge River) (Figure 1). Three of these rivers belong to the
Mediterranean lineage (ME; the Aude, the Doron and the Roya Rivers), whereas the two others
belong to the Atlantic lineage (AT; the One and the Seuge Rivers) (see Figure 1).
47
Figure 1.1: Map of the five river basins and sampling sites (black dots) used to test for genotyping success.
The Seuge and Ône Rivers are part of the Atlantic catchment, therefore naturally harboring AT trout,
whereas the Aude, Roya and Doron de Bozel Rivers are part of the Mediterranean catchment, naturally
populated with ME lineage. Maps were generated by authors on ArcGis and assembled using Inkscape.
The sampling sessions were performed in 2016, using a single-pass electrofishing approach from a
total of 79 sites (between 8 and 21 sites per river basin, Appendix S5), with an aim of sampling 30
individuals of brown trout per site. In total, we captured 1862 individuals (26 individuals per site
in average; see Appendix S5) from which a fin clip was taken (after Eugenol anesthesia), and kept
in 70% TE Ethanol for genotyping. All individuals were released alive to their site of capture. Fin
samples were sent to the LGC Genomics company for DNA extraction and multilocus genotyping
of the 192 SNPs markers using KASPAR® (Smith and Maughan, 2015). Genotyping success was
measured at the individual level, by the proportion of SNPs which were not genotyped (either
because not amplified or because the allele could not be read). Finally, these basins were tested for
patterns of isolation-by-distance, by using a mantel test with 1000 permutations on pairwise Fst
matrices (calculated with the pairwise.fst adegenet function) and riparian distances (measured in
meters with STARS ArcGis package).
48
Efficiency of the low-density SNPs array.
The low-density SNP array was further used for classic population genetic questions in order to
compare its efficiency with thirteen microsatellite markers previously used in brown trout
population genetic studies (e.g. Estoup et al., 1998; Grimholt et al., 2002; Slettan et al., 1995). A
total of 190 brown trout individuals were sampled in a small river basin (the Taurion River in the
Massif Central Mountains; Figure 1, ―test‖) in 2017 using electrofishing. Seven sites were sampled
with 21 to 30 individuals per site (Appendix S6). For each individual, a pelvic fin clip was taken
for genetic analyses. All individuals were released to their original sampling site. We additionally
sampled 30 individuals of domestic Atlantic brown trout from a local hatchery used for stocking
purposes (the Soueich trout hatchery), to quantify genetic admixture with wild populations from
the Taurion basin. Fin samples were sent to the LGC Genomics Company for DNA extraction and
for multilocus genotyping to 192 SNPs markers using the KASPAR® (Smith and Maughan,
2015). Note that 30 SNP markers from the initial 192 SNP panel did not amplify (see the Results),
and were hence replaced by 30 other SNPs from the 245 SNPs filtered (see methods) to improve
the SNP array (see Appendix S2 for details). Additionally, all individuals were genotyped at
thirteen microsatellites assembled in PCR multiplexes (see Appendix S7, and Saint-Pé et al., 2018
for details). We tested whether the selected SNPs were likely influenced by selection using the Fst
outlier detection method implemented in the fsthet R package (Flanagan and Jones, 2017), in
which outlier values of FST can be identified in a plot of FST vs. heterozygosity (Beaumont and
Nichols, 1996).
From both SNP and microsatellite datasets, we removed individuals with more than a third of
missing data, and kept only individuals for which we had both SNPs and microsatellites
genotypes. We first compared genetic admixture between wild and captive-bred strains using
STRUCTURE 2.3.1 (Pritchard et al., 2000) with the admixture model and the correlated allele
frequency model, without prior population information. Twenty runs assuming two clusters (K=2,
in order to discriminate between wild and captive-bred individuals, see (Saint-Pé et al., 2018))
were performed with a burn-in period of 200,000 and 200,000 subsequent MCMC repetitions. The
ten best runs (highest LnP(D) values) were compiled using CLUMPP (Jakobsson and Rosenberg,
2007) to obtain final averaged individual Q-values. Individuals were assigned to one of the two
clusters with the greatest Q-value, provided that value exceeded 0.7 (as in (Saint-Pé et al., 2018)).
49
Individuals with intermediate Q-values were considered genetically admixed individuals between
hatchery and wild strains. Individual assignment Q-values to the cluster containing all Soueich
hatchery individuals (i.e. degree of assignment to the ―captive-bred‖ cluster = individual level of
―hatchery ancestry‖) were compared between SNPs and microsatellites using a Spearman
correlation test, as admixed ancestry was not normally distributed.
We then compared population differentiation assessment between markers by calculating pairwise
Fst between sites using the adegenet R package and mantel R function. Finally, we compared the
informativeness of both sets of markers for population structure by calculating the informativeness
for assignment (In; (Rosenberg et al., 2003)). A higher index indicates a higher informativeness of
the set of markers. It was calculated with R as follows: for i = 1, 2,..., K populations and m = 1, 2,
..., L loci, with K ≥ 2 and L ≥ 1. Locus m has alleles j = 1, 2, ..., N(m). The average frequency of
allele j at locus m across the K populations is defined as = , where is the
relative frequency for allele j of locus m in population i. The informativeness is defined as:
= .
Results
Development of the large SNPs resource and characteristics of the low-density SNP array
After applying filters, we identified a RAD data set of 12,204 sequences each containing one or
two SNPs that met our specifications and that are made readily available to the scientific
community (see Appendix S1). The number of RADs per linkage groups (LGs) varies between
137 (LG 33) and 563 (LG 6). These RAD tags are spaced by 0.119 cM (+/- 0.039) in average.
They were found in all LGs and they cover the linkage map relatively evenly, although there were
large gaps on LGs 11 (top), 12 (top and bottom), 27 (bottom), 32 (top), 33 (bottom), 37 (bottom),
and 39 (top) (Figure 2A). One estimate of the local recombination rate for each SNP (i.e. one
index of the relative power of markers for LD-based mapping approaches) is provided in
Appendices S1 and S2.
50
The 245 SNPs selected from this set of RADs were spread over all linkage group so as to cover the
linkage map as homogeneously as possible (Figure 2B). The final low-density 182 SNP array, to
which we added 10 SNPs previously developed (see the Methods) is presented in Appendix S2.
Figure 1.2: Positions on the brown trout linkage map of the 12204 RADs (A; containing one or two SNPs
with MAF ≥ 5% based on all individuals (AT + ME) and positions between the first or last 30 bp of the
RAD with no undetermined nucleotides), and of the 245 SNPs (B; informative for at least the AT and ME
lineages, positions between the first or last 50 bp of RADs (primer design zones), and spaced by at least 3.5
cM)
51
Genotyping success of the low-density SNPs array
Thirty of the 192 SNPs (among which a mitochondrial marker) that were initially genotyped in the
five watersheds did not amplify, suggesting primers failure to properly bind their target DNA site.
For the loci that successfully amplified, the overall genotyping success was high, with less than
1% of missing data per individual, irrespective of the river basin and the lineage (Table 1).
Nonetheless, the number of polymorphic SNPs greatly varied from one river to another, ranging
from 91 (for the Seuge River) to 160 (for the Aude River) out of the 162 SNPs that were
successfully genotyped in that portion of our work (Table 1). The isolation-by-distance patterns
were significant (pairwise Fst significantly correlated to pairwise riparian distance) for the Seuge
River (rMantel = 0.55, p-value <0.001) and the Roya River (rMantel = 0.65, p-value = 0.01). Fst and
riparian distance were also positively although non-significantly correlated for the Aude River
(rMantel = 0.19, p-value = 0.15), the One River (rMantel = 0.28, p-value = 0.07) and the Doron River
(rMantel = 0.02, p-value = 0.36) (see Appendix S3 for the plots).
River basin (lineage)
Number of polymorphic
loci
Average (±SD) number of non-genotyped loci per individual on
polymorphic loci Min-max
He on polymorphic
loci
Basin surface (Km2)
Aude (Mediteranean) 160/162
1.04 (±1.68) 0-24 0.16
240
Doron de Bozel (Mediteranean) 153/162
0.71 (±0.99) 0-8 0.30
180
One (Atlantic) 119/162
0.87 (±1.23) 0-13 0.14
155
Roya (Mediteranean) 158/162
1.36 (±4.51) 0-63 0.30
360
Seuge (Atlantic) 91/162
0.66 (±1.06) 0-15 0.09
90
Average 136 0.95 0.20
205
Table 1.1: Number of polymorphic loci, number of non-genotyped loci per individual considering
polymorphic loci (average, min and max), expected heterozygosity (He) on polymorphic loci in each of the
five river basins, and surface drained by the river basin (upstream from the lowest sampling site).
52
Efficiency of the low-density SNPs array
Among the 30 replaced SNPs for the empirical test in the Taurion River, 10 failed at the
amplification step (see Appendix S2). This resulted in a set of 182 markers that successfully
amplified, although only 92 of them were polymorphic in this river (among which only one
mitochondrial marker of the 10 INRA and mitochondrial added markers). After removing
individuals with more than a third of missing data, and keeping only individuals for which we had
both SNPs and microsatellite data, our final dataset comprised 197 individuals (167 individuals
from the Taurion basin and 30 from the Soueich trout farm). Expected heterozygosity ranged from
0.18 to 0.22 for SNPs and from 0.58 to 0.73 for microsatellites. Fst ranged from 0.022 to 0.070 for
SNPs and from 0.019 to 0.040 for microsatellites (details for each site are shown in Table 2). We
found one marker that was a significant Fst outlier, suggesting it may be influenced by selection
(ID 295415; Fst>>Expected heterozygosity), but we decided to keep it in our analyses (see the
Discussion).
SNPs Microsatellites
Site N Ho He Fis Fst Ar Ho He Fis Fst Ar
BEA-Rau 27 0.18 ± 0.19 0.18 ± 0.18 0.011 0.070 1.58 0.44 ± 0.20 0.58 ± 0.25 0.226 0.04 4.42
PON-Rau 12 0.22 ± 0.23 0.21 ± 0.20 -0.038 0.030 1.66 0.67 ± 0.19 0.70 ± 0.17 0.027 0.022 5.63
THA-Bar 26 0.22± 0.23 0.22 ± 0.20 -0.025 0.022 1.65 0.65 ± 0.17 0.67 ± 0.18 0.015 0.017 5.85
THA-Tcc 30 0.19 ± 0.20 0.20 ± 0.18 0.028 0.027 1.69 0.61 ± 0.18 0.69 ± 0.17 0.107 0.019 6.18
THA-Usi 29 0.22 ± 0.22 0.21 ± 0.19 -0.023 0.025 1.67 0.68 ± 0.12 0.73 ± 0.16 0.057 0.019 6.79
THA-Vig 21 0.21 ± 0.21 0.22 ± 0.20 0.056 0.028 1.68 0.64 ± 0.17 0.72 ± 0.12 0.125 0.024 6.40
VIG-Tex 22 0.21 ± 0.21 0.21 ± 0.19 -0.004 0.031 1.66 0.69 ± 0.14 0.70 ± 0.15 0.033 0.02 5.91
PIS-Sou 0.34 ± 0.18 0.33 ± 0.16 -0.008 0.153 1.96 0.80 ± 0.11 0.84 ± 0.09 0.052 0.070 8.14
Table 1.2: For each site, sample size (“N”), mean expected heterozygosity over all loci (“He”) and
standard deviation between loci, mean observed heterozygosity over all loci (“Ho”) and standard deviation
between loci, mean allelic richness computed using a rarefaction approach over all loci (“Ar”), mean Fis
over all loci (“Fis”), mean Fst over all loci (i.e. uniqueness at the site level; Fst = 1-HeSite/HeTotal).
Individual inferences of hatchery ancestry measured with either SNPs or microsatellites are
presented in Figure 3 in the form of individual barplots. The distribution of captive-bred ancestry
was bimodal, meaning that most individuals were either purely wild or captive-bred, with
relatively low numbers of admixed genotypes (Figure 3, Figure 4A). Levels of hatchery ancestry
53
were significantly and moderately correlated (rSpearman = 0.60, d.f. = 193, p < 0.001; Figure 4A),
although for some individuals, there was a discrepancy between markers with one of the two
marker types detecting introgressed fish while the other marker type assigned them as pure wild
fish (Figure 3, Figure 4A). Pairwise Fst values are presented for both SNPs and microsatellites in
Appendix S4. They ranged from 0.010 to 0.088 for SNPs (mean = 0.033 ± 0.025) and from 0.010
to 0.053 for microsatellites (mean = 0.025 ± 0.013). Pairwise Fst values between sites assessed
with SNPs and microsatellites were strongly correlated (rMantel = 0.92, p = 0.001 based on 1000
permutations), and SNPs had higher values (Figure 4B, the regression coefficient is significantly
higher than the 1:1 line (dotted line on figure) since its 95% CI is 1.46-2.18).
Figure 1.3: Structure barplots of assignment to the wild (grey) and the captive-bred (black) clusters, using
both SNPs (A) and microsatellites (B)
54
Figure 1.4: Plot of individual hatchery ancestry proportion measured with microsatellites against that
measured with SNPs (A) and pairwise Fst between sites measured with microsatellites against pairwise Fst
between sites measured with SNPs (B). Black lines represent the linear regression with its confidence
interval; spotted lines represent the 1:1 line.
Finally, within the Taurion River, the 92 SNPs displayed a slightly lower informativeness (I) than
the 13 microsatellites (2.08 vs. 2.48; Figure 5). We found that the informativeness of these 92
SNPs is actually equivalent to that of 10 of the microsatellites (Figure 5). Based on the equation
linking the number of our SNPs and informativeness (I = 0.023*Nsnps+0.0038; r2 = 0.98, p-value
< 0.001), we extrapolated that 108 SNPs are required to be equivalent to the panel of 13
microsatellites in terms of informativeness for individual assignment.
55
Figure 1.5: Plot of informativeness against number of SNPs. 107 SNPs would be equivalent to the 13
microsatellites in terms of informativeness.
Discussion
The SNP panel developed here was shown to be efficient to study the population genetics of
Atlantic and Mediterranean brown trout lineages from Western Europe, which gathered a huge
number of studies in the past decades (Poteaux, 1999; Aurelle and Berrebi, 2001; Almodóvar et
al., 2012). We provided a panel of 12,204 RADs which are relatively evenly spread across the
whole genome, and, from a sample of this panel, we proved its efficiency in terms of genotyping
success, and measuring patterns of isolation-by-distance. We also proved that these SNPs
successfully detect population structure, which opens new insights for many applications with
great potential compared to commonly used markers. These advantages are multiple, going from
lower error rates and a simple mutation model with low homoplasy (Morin and Mccarthy, 2007) to
usability on poor quality samples. In terms of costs, SNPs and microsatellites are roughly similar,
however in terms of time efficiency, SNPs are highly advantageous: samples are directly sent to
56
the genotyping platform, and within a month, data is ready to be analyzed. Moreover, SNPs can be
easily reused in other studies, and are more powerful in detecting hybridization (McFarlane and
Pemberton, 2019).
Development of a large variation map for population genomic studies in brown trout
The 12,204 RAD-derived SNPs were relatively homogeneously distributed across the 40 linkage
groups of Salmo trutta (with 137 to 563 RADs per LG), and showed an average spacing of 0.119
cM. There are some gaps in coverage, which does not necessarily mean that information is
missing, they could be due to high recombination rates in these regions or to the positions of the
centromeres. Although these gaps exist, genome coverage is satisfying, and with the marker
density obtained, it should be sufficient for most genome-scale studies that need to tag a large
fraction of genomic variation through linkage disequilibrium. However, this marker density may
still be limited when a rapid decay of linkage disequilibrium leaves many genomic regions
unattainable, as it might be the case for some applications like the search for loci underlying
fitness in the wild (Kaeuffer et al., 2007). Therefore, an estimate of the local recombination rate
for each SNP as an index of the relative power of markers for LD-based mapping approaches is
provided in Appendices S1 and S2.
The proposed SNP resource only includes SNPs with a MAF higher than 0.05 (i.e. removing of
rare variants), a criterion that has been set for two main reasons. First, SNPs with very low MAF
can in many cases be genotyping errors (Ahn et al., 2009; Wang et al., 2008). Second, we chose a
relatively high MAF because this panel of SNPs is primarily designed for studying populations for
most of the species range. Therefore, SNPs that are discovered in the populations used to develop
the panel are more likely to be polymorphic in other populations from which they were not
developed if their MAF is high (Morin et al., 2009). Although filtering SNPs on their MAF could
lead to ascertainment bias (Helyar et al., 2011; Nielsen et al., 2009; Rosenblum and Novembre,
2007), we suggest this is not an issue in our case study because we applied the MAF criteria
regarding distinct glacial lineages, which limits marker choice bias. Moreover, for individually-
centered investigations such as population structure, kinship and individual assignment, SNPs
with higher MAF were shown to generally be the most powerful (Krawczak, 1999; Morin et al.,
2004; Ryman et al., 2006).
57
In the test panel, and for the Taurion river (test basin), one locus might be potentially affected by
selection. Outlier markers are usually removed before analyses for inferring neutral evolutionary
processes, such as genetic drift and gene flow (Beaumont and Nichols, 1996). In our analyses, we
decided not to remove it because in the case of detection of admixture between strains, it was
shown that these ancestry informative markers can increase accuracy for detecting differentiation
and assignment of individuals to populations (Funk et al., 2012; Gagnaire et al., 2015; Nielsen et
al., 2009). Removing loci displaying selection is thus up to the users of the resource, depending on
the aim of the study. For instance, if the aim is to determine if a population is at HWE, or to
quantify gene flow/inbreeding, or calculate effective population sizes, loci affected by selection
should be removed, whereas more individual-based questions do not necessarily require removing
these loci.
Genotyping success
Genotyping success was very high in all basins, and except for two basins, the number of
polymorphic SNPs among the 162 amplified (from the 192 set) was satisfactory (73 to 99 %),
confirming the potential versatility of this tool. This set of 162 SNPs already benefits to the field
of population genetics for the species as to our knowledge, most studies used less than 40 SNPs
(Pustovrh et al., 2012; Drywa et al., 2013; Sušnik Bajec et al., 2015). Therefore, the 12,204 RAD
panel is a promising tool for genome wide studies on brown trout. It would be of interest however
to further test this panel on other evolutionary lineages, or on populations which have been shown
to have diverged from the continental populations such as western Mediterranean populations
found in Iberian Peninsula (Cortey et al., 2004), Italian populations (Splendiani et al., 2016), and
Corsican populations (Berrebi et al., 2019), as well as on the other main lineages (e.g. Adriatic and
Danube lineages; Bernatchez, 2001) or remote populations inhabiting at the edges of the species‘
range (Iran: Hashemzadeh Segherloo et al., 2012; Osinov, 2009, Morocco: Tougard et al., 2018).
As a first approach we found this panel to be efficient for detecting patterns of isolation-by-
distance, although we had no other markers to compare with. Patterns of isolation-by-distance
were found to be significant only in the Seuge and the Roya Rivers, although there was a tendency
in the other three basins. Interestingly, it seems that even when the number of polymorphic SNPs
was low (Seuge River: 92 polymorphic loci), detecting a pattern of isolation-by-distance was still
possible. In other basins, we suggest that the strength of the relation between genetic and riparian
58
distance may also be affected by stocking events, or by characteristics of the watershed and
demographic histories. However, we did not investigate these issues further in the present
manuscript.
Tests of the low-density SNP array
Studies in which low-density SNP arrays equal or outperform a handful of microsatellites for
population structure and differentiation are common, particularly when sample sizes are large and
populations are strongly structured (Freamo et al., 2011; Gärke et al., 2012; Liu et al., 2005).
Although we found no literature on this particular aspect in brown trout, it has been shown in
Atlantic Salmon : genetic divergence, structuring and isolation-by-distance were assessed as
successfully using only 7 SNPs and 14 microsatellites, although genetic diversity estimates were
less concordant (Ryynanen et al., 2007). Twenty-six SNPs were also shown to be nearly as
efficient as 16 microsatellites for parentage assignment in this species (Rengmark et al., 2006),
which makes our set of SNPs appear very promising. The panel of 192 SNPs tested here
performed well in terms of detection of admixture and population differentiation, although only 92
SNPs were polymorphic in the Taurion basin. The low level of polymorphism is probably due to
the fact that the study scale is extremely low (less than 15 km between the two most extreme
sampling sites; Figure 1) and/or that the biogeographic area in which this river basin is situated has
historically low level of diversity (see below).
We found similar introgression level when measured with SNPs compared to microsatellites
(Figure 4A), and higher pairwise Fst values between sites with SNPs, suggesting that SNPs have –
at least - a similar discriminatory and assignment power. However, the low number of
polymorphic SNPs in this panel (92 polymorphic SNPs) compared to other SNP-microsatellite
comparison case studies (Gärke et al., 2012; Hauser et al., 2011; Kaiser et al., 2017) lowered its‘
informativeness : it was outperformed by the 13 microsatellites, and we found that it would
actually require 107 SNPs to be equivalent to microsatellites in terms of informativeness for
assignment. However, the advantage of SNPs may here not be accounted for when calculating
informativeness. Indeed, they enable to better detect introgression and admixture compared to
microsatellites, and show that individuals considered as « pure » with microsatellites may in fact
be introgressed (Haasl and Payseur, 2011). This difference in individual admixture proportions
59
calculated with microsatellites and SNPs might results from the fact that our panel is characterized
by an even repartition of SNPs along the genome of the brown trout, which is expected to improve
the global assessment of genome-wide admixture proportions (Leitwein et al., 2018). Hence, the
strong advantage of a SNP panel of this type is that it ensures a better representativeness of the
entire genome of the brown trout. Moreover, although informativeness is lower, the 92 SNPs still
give sufficient information on admixture and differentiation, highly correlated with that given by
the microsatellites: the trade-off between cost and power must also be taken into account in regard
of the questions asked and the means of the user.
For other empirical case studies, such as the five river basins on which genotyping success was
measured, and in which individuals are variable on more loci (except for the Seuge river, other
basins showed 119 to 160 polymorphic loci, Table 1), we can expect higher informativeness. We
even expect SNPs to equal or outperform microsatellites in terms of informativeness for these
basins (indeed, around 107 SNPs should be equivalent to the panel of 13 microsatellites, as
pairwise Fst values were higher for SNPs than microsatellites), and with the advantages of SNPs
mentioned previously. This is particularly true in Mediterranean rivers, probably because they
contain both domestic Atlantic and natural Mediterranean ancestries. Indeed, these SNPs were
discovered using a mix of Mediterranean and Atlantic individuals, and are therefore more likely to
be polymorphic if both lineages are present. Additionally, the mitochondrial and INRA markers
were also developed to have a fixed allele in each lineage, explaining why in the Taurion for
instance, in which only the Atlantic lineage is present, they were not polymorphic (except for one).
We make this SNP panel freely available as a resource. As it contains many more untested SNPs,
future users will be able to choose the number, density, and position of markers in the linkage
map, and considering the local recombination rate around each SNP in order to adjust their own
panel to their objectives. This should also be considered for the Massif Central Rivers (in our case
Taurion and Seuge Rivers), in which the number of polymorphic loci is lower than in the other
basins studied here, probably because of the past demographic histories (colonization,
connectivity, stocking, bottlenecks, population sizes and habitat). However, as this subpanel was
tested on only 7 populations (190 individuals), we first suggest that increasing the sample size
should increase statistical power better than increasing the number of SNPs, especially when Fst is
low (< 0.01; Kalinowski, 2005; Morin et al., 2009).
60
Conclusion
The SNP panel presented here appears as a novel tool to study diverse aspects of population
genetics in the brown trout. The possibility to genotype many loci in a fast and affordable way will
open many perspectives. It opens new insights into the species life history, with many potential
applications both for fundamental population genetics, conservation and management questions,
but also for more biological questions such as mapping of quantitative trait loci, or investigating
links between genetic and environmental divergence. This resource has the potential to offer high
flexibility for many possible applications, outperforming previously used markers in many ways:
genome coverage and ancestry detection for instance, but also in terms of cost and efficiency to
obtain individual genotypes. We hope that it will be useful to the population geneticists‘
community working on brown trout and call for future studies across the species‘ range.
Declarations
Ethics approval and consent to participate
The Direction Départmentale des Territoires et de la Mer of departments gave their ethical
approval for the field sessions and trout capture.
Availability of data and material
SNPs sequences and information are available on Figshare.
DOI: 10.6084/m9.figshare.8174708
Competing interests
We have no competing interests and we declare no conflict of interests.
Funding
This work was funded by the French Agency for Biodiversity (n° 136406), Electricité De France
(n°8619-5920011137). Two of the authors are attached to the funding body (Tissot Laurence,
61
EDF; Poulet Nicolas, AFB), and have contributed to the design of the study, the collection,
analysis, and interpretation of data, and in writing the manuscript.
Authors' contributions
KSP, SB, NP and LT designed the experiment and coordinated the study; KSP, ML, BG, LT, NP,
GM, JML, PB, and SB conducted sampling; ML, PAG, BG and PB generated the variation map
used for developing the SNP panel; KSP, ML, GM and SB carried out the experimental lab work;
KSP and ML ran the statistical analyses; KSP, ML, PAG, GB, LT, NP, and SB interpreted the
data; KSP, ML PAG, and SB wrote the first draft of the manuscript; BG, LT, NP, GM, JML and
PB read, commented and corrected the initial draft, and all authors gave final approval for
publication.
Acknowledgments
We warmly thank the AAPPMA of Luchon (J. Lérimé, D. Estrade and colleagues) and of Hérault
(E. Ravel and colleagues), ECOGEA (P. Baran and colleagues) and members of EDF (A. Barillier,
V. Gouraud and colleagues) for field sampling support. This work was undertaken at SETE, which
forms part of the ―Laboratoire d‘Excellence‖ (LABEX) entitled TULIP (ANR-10-LABX-41), and
M. Leitwein was supported by a grant from LABEX CeMEB (Centre Méditerranéen de
l‘Environnement et de la Biodiversité, Montpellier), using genotyping and sequencing facilities of
ISEM (Institut des Sciences de l‘Evolution-Montpellier)
References
Ahn, K., Gordon, D., Finch, S.J., 2009. Increase of rejection rate in case-control studies with the differential
genotyping error rates. Stat. Appl. Genet. Mol. Biol. 8, 1–9. https://doi.org/10.2202/1544-
6115.1429
Almodóvar, A., Nicola, G.G., Ayllón, D., Elvira, B., 2012. Global warming threatens the persistence of
Mediterranean brown trout. Glob. Change Biol. 18, 1549–1560. https://doi.org/10.1111/j.1365-
2486.2011.02608.x
Altukhov, I.P., Salmenkova, E.A., Omelʹchenko, V.T., 2000. Salmonid fishes: population biology, genetics,
and management, Fish and aquatic resources series. Blackwell Science, Oxford [England] ;
Malden, MA, USA.
Apostolidis, A.P., Triantaphyllidis, C., Kouvatsi, A., Economidis, P.S., 1997. Mitochondrial DNA sequence
variation and phylogeography among Salmo trutta L. (Greek brown trout) populations. Mol. Ecol.
6, 531–542. https://doi.org/10.1046/j.1365-294X.1997.d01-176.x
Araki, H., Schmid, C., 2010. Is hatchery stocking a help or harm? Aquaculture 308, S2–S11.
https://doi.org/10.1016/j.aquaculture.2010.05.036
62
Araki, H., Waples, R.S., Ardren, W.R., Cooper, B., Blouin, M.S., 2007. Effective population size of
steelhead trout: influence of variance in reproductive success, hatchery programs, and genetic
compensation between life-history forms: effective population size of steelhead trout. Mol. Ecol.
16, 953–966. https://doi.org/10.1111/j.1365-294X.2006.03206.x
Aurelle, D., Berrebi, P., 2001. Genetic structure of brown trout (Salmo trutta, L.) populations from south-
western France: data from mitochondrial control region variability. Mol. Ecol. 10, 1551–1561.
https://doi.org/10.1046/j.1365-294X.2001.01293.x
Bardakci, F., Degerli, N., Ozdemir, O., Basibuyuk, H.H., 2006. Phylogeography of the Turkish brown trout
Salmo trutta L.: mitochondrial DNA PCR-RFLP variation. J. Fish Biol. 68, 36–55.
https://doi.org/10.1111/j.0022-1112.2006.00948.x
Beaumont, M.A., Nichols, R.A., 1996. Evaluating loci for use in the genetic analysis of population
structure. Proc. R. Soc. Lond. B Biol. Sci. 263, 1619–1626. https://doi.org/10.1098/rspb.1996.0237
Bernatchez, L., 2001. The evolutionary history of brown trout (Salmo trutta L.) inferred from
phylogeographic, nested clade, and mismatch analyses of mitochondrial dna variation. Evolution
55, 351. https://doi.org/10.1554/0014-3820(2001)055[0351:TEHOBT]2.0.CO;2
Bernatchez, L., Guyomard, R., Bonhomme, F., 1992. DNA sequence variation of the mitochondrial control
region among geographically and morphologically remote European brown trout Saltmo trutta
populations. Mol. Ecol. 1, 161–173. https://doi.org/10.1111/j.1365-294X.1992.tb00172.x
Berrebi, P., 2015. Three brown trout Salmo trutta lineages in Corsica described through allozyme variation:
salmo trutta genetics in corsica. J. Fish Biol. 86, 60–73. https://doi.org/10.1111/jfb.12534
Berrebi, P., Caputo Barucchi, V., Splendiani, A., Muracciole, S., Sabatini, A., Palmas, F., Tougard, C.,
Arculeo, M., Marić, S., 2019. Brown trout (Salmo trutta L.) high genetic diversity around the
Tyrrhenian Sea as revealed by nuclear and mitochondrial markers. Hydrobiologia 826, 209–231.
https://doi.org/10.1007/s10750-018-3734-5
Berrebi, P., Poteaux, C., Fissier, M., Cattaneo-Berrebi, G., 2000. Stocking impact and allozyme diversity in
brown trout from Mediterranean southern France. J. Fish Biol. 56, 949–960.
https://doi.org/10.1111/j.1095-8649.2000.tb00884.x
Bohling, J., Haffray, P., Berrebi, P., 2016. Genetic diversity and population structure of domestic brown
trout (Salmo trutta) in France. Aquaculture 462, 1–9.
https://doi.org/10.1016/j.aquaculture.2016.04.013
Bohling, J.H., 2016. Strategies to address the conservation threats posed by hybridization and genetic
introgression. Biol. Conserv. 203, 321–327. https://doi.org/10.1016/j.biocon.2016.10.011
Bourret, V., Kent, M.P., Primmer, C.R., Vasemägi, A., Karlsson, S., Hindar, K., McGinnity, P., Verspoor,
E., Bernatchez, L., Lien, S., 2013. SNP-array reveals genome-wide patterns of geographical and
potential adaptive divergence across the natural range of Atlantic salmon (Salmo salar). Mol. Ecol.
22, 532–551. https://doi.org/10.1111/mec.12003
Bradbury, I.R., Hamilton, L.C., Dempson, B., Robertson, M.J., Bourret, V., Bernatchez, L., Verspoor, E.,
2015. Transatlantic secondary contact in Atlantic salmon, comparing microsatellites, a single
nucleotide polymorphism array and restriction-site associated DNA sequencing for the resolution
of complex spatial structure. Mol. Ecol. 24, 5130–5144. https://doi.org/10.1111/mec.13395
Brumfield, R.T., Beerli, P., Nickerson, D.A., Edwards, S.V., 2003. The utility of single nucleotide
polymorphisms in inferences of population history. Trends Ecol. Evol. 18, 249–256.
https://doi.org/10.1016/S0169-5347(03)00018-1
Butler, J.R.A., Radford, A., Riddington, G., Laughton, R., 2009. Evaluating an ecosystem service provided
by Atlantic salmon, sea trout and other fish species in the River Spey, Scotland: The economic
impact of recreational rod fisheries. Fish. Res. 96, 259–266.
https://doi.org/10.1016/j.fishres.2008.12.006
Carlsson, J., Olsen, K.H., Nilsson, J., Overli, O., Stabell, O.B., 1999. Microsatellites reveal fine-scale
genetic structure in stream-living brown trout. J. Fish Biol. 55, 1290–1303.
https://doi.org/10.1111/j.1095-8649.1999.tb02076.x
63
Caudron, A., Champigneulle, A., Guyomard, R., 2006. Assessment of restocking as a strategy for
rehabilitating a native population of brown trout Salmo trutta L. in a fast-flowing mountain stream
in the northern French Alps. J. Fish Biol. 69, 127–139. https://doi.org/10.1111/j.1095-
8649.2006.01156.x
Christie, M.R., Marine, M.L., French, R.A., Waples, R.S., Blouin, M.S., 2012. Effective size of a wild
salmonid population is greatly reduced by hatchery supplementation. Heredity 109, 254–260.
https://doi.org/10.1038/hdy.2012.39
Cortey, M., Garcia-Marin, J.-L., 2002. Evidence for phylogeographically informative sequence variation in
the mitochondrial control region of Atlantic brown trout. J. Fish Biol. 60, 1058–1063.
https://doi.org/10.1111/j.1095-8649.2002.tb02429.x
Cortey, M., Pla, C., García-Marín, J.-L., 2004. Historical biogeography of Mediterranean trout. Mol.
Phylogenet. Evol. 33, 831–844. https://doi.org/10.1016/j.ympev.2004.08.012
Danecek, P., Auton, A., Abecasis, G., Albers, C.A., Banks, E., DePristo, M.A., Handsaker, R.E., Lunter,
G., Marth, G.T., Sherry, S.T., McVean, G., Durbin, R., 1000 Genomes Project Analysis Group,
2011. The variant call format and VCFtools. Bioinformatics 27, 2156–2158.
https://doi.org/10.1093/bioinformatics/btr330
Davey, J.W., Hohenlohe, P.A., Etter, P.D., Boone, J.Q., Catchen, J.M., Blaxter, M.L., 2011. Genome-wide
genetic marker discovery and genotyping using next-generation sequencing. Nat. Rev. Genet. 12,
499–510. https://doi.org/10.1038/nrg3012
Dominik, S., Henshall, J.M., Kube, P.D., King, H., Lien, S., Kent, M.P., Elliott, N.G., 2010. Evaluation of
an Atlantic salmon SNP chip as a genomic tool for the application in a Tasmanian Atlantic salmon
(Salmo salar) breeding population. Aquaculture 308, S56–S61.
https://doi.org/10.1016/j.aquaculture.2010.05.038
Drywa, A., Poćwierz-Kotus, A., Wąs, A., Dobosz, S., Kent, M.P., Lien, S., Bernaś, R., Wenne, R., 2013.
Genotyping of two populations of Southern Baltic Sea trout Salmo trutta m. trutta using an Atlantic
salmon derived SNP-array. Mar. Genomics 9, 25–32. https://doi.org/10.1016/j.margen.2012.08.001
Elliott, J.M., 1994. Quantitative ecology and the brown trout, Oxford series in ecology and evolution.
Oxford University Press, Oxford ; New York.
Estoup, A., Rousset, F., Michalakis, Y., Cornuet, J.-M., Adriamanga, M., Guyomard, R., 1998.
Comparative analysis of microsatellite and allozyme markers: a case study investigating
microgeographic differentiation in brown trout (Salmo trutta). Mol. Ecol. 7, 339–353.
https://doi.org/10.1046/j.1365-294X.1998.00362.x
Etter, P.D., Preston, J.L., Bassham, S., Cresko, W.A., Johnson, E.A., 2011. Local De Novo Assembly of
RAD Paired-End Contigs Using Short Sequencing Reads. PLoS ONE 6, e18561.
https://doi.org/10.1371/journal.pone.0018561
Flanagan, S.P., Jones, A.G., 2017. Constraints on the FST–Heterozygosity Outlier Approach. J. Hered. 108,
561–573. https://doi.org/10.1093/jhered/esx048
Freamo, H., O‘Reilly, P., Berg, P.R., Lien, S., Boulding, E.G., 2011. Outlier SNPs show more genetic
structure between two Bay of Fundy metapopulations of Atlantic salmon than do neutral SNPs:
SNP genotyping and applications. Mol. Ecol. Resour. 11, 254–267. https://doi.org/10.1111/j.1755-
0998.2010.02952.x
Funk, W.C., McKay, J.K., Hohenlohe, P.A., Allendorf, F.W., 2012. Harnessing genomics for delineating
conservation units. Trends Ecol. Evol. 27, 489–496. https://doi.org/10.1016/j.tree.2012.05.012
Gagnaire, P.-A., Broquet, T., Aurelle, D., Viard, F., Souissi, A., Bonhomme, F., Arnaud-Haond, S., Bierne,
N., 2015. Using neutral, selected, and hitchhiker loci to assess connectivity of marine populations
in the genomic era. Evol. Appl. 8, 769–786. https://doi.org/10.1111/eva.12288
Gärke, C., Ytournel, F., Bed‘hom, B., Gut, I., Lathrop, M., Weigend, S., Simianer, H., 2012. Comparison of
SNPs and microsatellites for assessing the genetic structure of chicken populations: Population
differentiation: SNPs vs SSRs. Anim. Genet. 43, 419–428. https://doi.org/10.1111/j.1365-
2052.2011.02284.x
64
Giuffra, E., Bernatchez, L., Guyomard, R., 1994. Mitochondrial control region and protein coding genes
sequence variation among phenotypic forms of brown trout Salmo trutta from northern Italy. Mol.
Ecol. 3, 161–171.
Grimholt, U., Drabløs, F., Jørgensen, S., Høyheim, B., Stet, R., 2002. The major histocompatibility class I
locus in Atlantic salmon (Salmo salar L.): polymorphism, linkage analysis and protein modelling.
Immunogenetics 54, 570–581. https://doi.org/10.1007/s00251-002-0499-8
Haasl, R.J., Payseur, B.A., 2011. Multi-locus inference of population structure: a comparison between
single nucleotide polymorphisms and microsatellites. Heredity 106, 158–171.
https://doi.org/10.1038/hdy.2010.21
Hansen, M.M., 2002. Estimating the long-term effects of stocking domesticated trout into wild brown trout
(Salmo trutta) populations: an approach using microsatellite DNA analysis of historical and
contemporary samples. Mol. Ecol. 11, 1003–1015. https://doi.org/10.1046/j.1365-
294X.2002.01495.x
Harrang, E., Guyomard, R., Caudron, A., 2015. Ready-to-use diagnostic molecular markers for assessment
and monitoring of hybridization between native and non-native domesticated brown trout (Salmo
trutta) : Case studies in French Alpine Streams. Adv. Popul. Ecol. Stream Salmon. – IV Girona
Spain.
Hashemzadeh Segherloo, I., Farahmand, H., Abdoli, A., Bernatchez, L., Primmer, C.R., Swatdipong, A.,
Karami, M., Khalili, B., 2012. Phylogenetic status of brown trout Salmo trutta populations in five
rivers from the southern Caspian Sea and two inland lake basins, Iran: a morphogenetic approach.
J. Fish Biol. 81, 1479–1500. https://doi.org/10.1111/j.1095-8649.2012.03428.x
Hauser, L., Baird, M., Hilborn, R., Seeb, L.W., Seeb, J.E., 2011. An empirical comparison of SNPs and
microsatellites for parentage and kinship assignment in a wild sockeye salmon (Oncorhynchus
nerka) population: Analytical approaches. Mol. Ecol. Resour. 11, 150–161.
https://doi.org/10.1111/j.1755-0998.2010.02961.x
Helyar, S.J., Hemmer-Hansen, J., Bekkevold, D., Taylor, M.I., Ogden, R., Limborg, M.T., Cariani, A.,
Maes, G.E., Diopere, E., Carvalho, G.R., Nielsen, E.E., 2011. Application of SNPs for population
genetics of nonmodel organisms: new opportunities and challenges: Analytical approaches. Mol.
Ecol. Resour. 11, 123–136. https://doi.org/10.1111/j.1755-0998.2010.02943.x
Jakobsson, M., Rosenberg, N.A., 2007. CLUMPP: a cluster matching and permutation program for dealing
with label switching and multimodality in analysis of population structure. Bioinformatics 23,
1801–1806. https://doi.org/10.1093/bioinformatics/btm233
Kaeuffer, R., Réale, D., Coltman, D.W., Pontier, D., 2007. Detecting population structure using
STRUCTURE software: effect of background linkage disequilibrium. Heredity 99, 374–380.
https://doi.org/10.1038/sj.hdy.6801010
Kaiser, S.A., Taylor, S.A., Chen, N., Sillett, T.S., Bondra, E.R., Webster, M.S., 2017. A comparative
assessment of SNP and microsatellite markers for assigning parentage in a socially monogamous
bird. Mol. Ecol. Resour. 17, 183–193. https://doi.org/10.1111/1755-0998.12589
Kalinowski, S.T., 2005. Do polymorphic loci require large sample sizes to estimate genetic distances?
Heredity 94, 33–36. https://doi.org/10.1038/sj.hdy.6800548
Karlsson, S., Moen, T., Lien, S., Glover, K.A., Hindar, K., 2011. Generic genetic differences between
farmed and wild Atlantic salmon identified from a 7K SNP-chip: SNP genotyping and applications.
Mol. Ecol. Resour. 11, 247–253. https://doi.org/10.1111/j.1755-0998.2010.02959.x
Klemetsen, A., Amundsen, P.-A., Dempson, J.B., Jonsson, B., Jonsson, N., O‘Connell, M.F., Mortensen,
E., 2003. Atlantic salmon Salmo salar L., brown trout Salmo trutta L. and Arctic charr Salvelinus
alpinus (L.): a review of aspects of their life histories. Ecol. Freshw. Fish 12, 1–59.
https://doi.org/10.1034/j.1600-0633.2003.00010.x
Kochzius, M., Nölte, M., Weber, H., Silkenbeumer, N., Hjörleifsdottir, S., Hreggvidsson, G.O.,
Marteinsson, V., Kappel, K., Planes, S., Tinti, F., Magoulas, A., Garcia Vazquez, E., Turan, C.,
65
Hervet, C., Campo Falgueras, D., Antoniou, A., Landi, M., Blohm, D., 2008. DNA Microarrays for
Identifying Fishes. Mar. Biotechnol. 10, 207–217. https://doi.org/10.1007/s10126-007-9068-3
Krawczak, M., 1999. Informativity assessment for biallelic single nucleotide polymorphisms.
Electrophoresis 20, 1676–1681. https://doi.org/10.1002/(SICI)1522-
2683(19990101)20:8<1676::AID-ELPS1676>3.0.CO;2-D
Laikre, L., Antunes, A., Concerted Action on Identification, M. and E. ofGenetic R. in the B.T. (Salmo
trutta) (projekt), 1999. Conservation genetic management of brown trout (Salmo trutta) in Europe.
Danmarks Fiskeriundersøgelser, Afd. for Ferskvandsfiskeri, Silkeborg.
Lamaze, F.C., Sauvage, C., Marie, A., Garant, D., Bernatchez, L., 2012. Dynamics of introgressive
hybridization assessed by SNP population genomics of coding genes in stocked brook charr
(Salvelinus fontinalis): SNP POPULATION GENOMICS IN BROOK CHARR. Mol. Ecol. 21,
2877–2895. https://doi.org/10.1111/j.1365-294X.2012.05579.x
Leitwein, M., Gagnaire, P.-A., Desmarais, E., Berrebi, P., GUINAND, B., 2018. Genomic consequences of
a recent three-way admixture in supplemented wild brown trout populations revealed by ancestry
tracts. bioRxiv. https://doi.org/10.1101/302380
Leitwein, M., Gagnaire, P.-A., Desmarais, E., Guendouz, S., Rohmer, M., Berrebi, P., Guinand, B., 2016.
Genome-wide nucleotide diversity of hatchery-reared Atlantic and Mediterranean strains of brown
trout Salmo trutta compared to wild Mediterranean populations: genome-wide nucleotide diversity
of s. trutta. J. Fish Biol. 89, 2717–2734. https://doi.org/10.1111/jfb.13131
Leitwein, M., Guinand, B., Pouzadoux, J., Desmarais, E., Berrebi, P., Gagnaire, P.-A., 2017. A Dense
Brown Trout (Salmo trutta) Linkage map reveals recent chromosomal rearrangements in the Salmo
genus and the impact of selection on linked neutral diversity. G3amp58 GenesGenomesGenetics 7,
1365–1376. https://doi.org/10.1534/g3.116.038497
Lien, S., Gidskehaug, L., Moen, T., Hayes, B.J., Berg, P.R., Davidson, W.S., Omholt, S.W., Kent, M.P.,
2011. A dense SNP-based linkage map for Atlantic salmon (Salmo salar) reveals extended
chromosome homeologies and striking differences in sex-specific recombination patterns. BMC
Genomics 12. https://doi.org/10.1186/1471-2164-12-615
Linløkken, A.N., Haugen, T.O., Kent, M.P., Lien, S., 2017. Genetic differences between wild and hatchery-
bred brown trout (Salmo trutta L.) in single nucleotide polymorphisms linked to selective traits.
Ecol. Evol. 7, 4963–4972. https://doi.org/10.1002/ece3.3070
Liu, N., Chen, L., Wang, S., Oh, C., Zhao, H., 2005. Comparison of single-nucleotide polymorphisms and
microsatellites in inference of population structure. BMC Genet. 6, S26.
https://doi.org/10.1186/1471-2156-6-S1-S26
Lobón-Cervía, J., Sanz, N. (Eds.), 2018. Brown trout: biology, ecology and management. Wiley, Hoboken,
NJ.
McFarlane, S.E., Pemberton, J.M., 2019. Detecting the true extent of Introgression during anthropogenic
hybridization. Trends Ecol. Evol. 34, 315–326. https://doi.org/10.1016/j.tree.2018.12.013
Mills, D., 1989. Conservation and management of brown trout, Salmo trutta, in Scotland: an historical
review and the future. Freshw. Biol. 21, 87–98. https://doi.org/10.1111/j.1365-
2427.1989.tb01350.x
Morin, P.A., Luikart, G., Wayne, R.K., the SNP workshop group, 2004. SNPs in ecology, evolution and
conservation. Trends Ecol. Evol. 19, 208–216. https://doi.org/10.1016/j.tree.2004.01.009
Morin, P.A., Martien, K.K., Taylor, B.L., 2009. Assessing statistical power of SNPs for population
structure and conservation studies. Mol. Ecol. Resour. 9, 66–73. https://doi.org/10.1111/j.1755-
0998.2008.02392.x
Morin, P.A., Mccarthy, M., 2007. Highly accurate SNP genotyping from historical and low-quality
samples. Mol. Ecol. Notes 7, 937–946. https://doi.org/10.1111/j.1471-8286.2007.01804.x
Nielsen, E.E., Hemmer-Hansen, J., Larsen, P.F., Bekkevold, D., 2009. Population genomics of marine
fishes: identifying adaptive variation in space and time. Mol. Ecol. 18, 3128–3150.
https://doi.org/10.1111/j.1365-294X.2009.04272.x
66
Osinov, A.G., 2009. Iranian samples of brown trout Salmo trutta or rainbow trout Oncorhynchus mykiss:
Comparison of evidence from allozyme and mitochondrial DNA control region sequence analysis.
J. Ichthyol. 49, 825–828. https://doi.org/10.1134/S0032945209090136
Paschou, P., Ziv, E., Burchard, E.G., Choudhry, S., Rodriguez-Cintron, W., Mahoney, M.W., Drineas, P.,
2007. PCA-Correlated SNPs for structure identification in worldwide human populations. PLoS
Genet. 3, e160. https://doi.org/10.1371/journal.pgen.0030160
Pelletier, F., Garant, D., Hendry, A.P., 2009. Eco-evolutionary dynamics. Philos. Trans. R. Soc. B Biol.
Sci. 364, 1483–1489. https://doi.org/10.1098/rstb.2009.0027
Poteaux, C., 1999. Microsatellite polymorphism and genetic impact of restocking in Mediterranean brown
trout (Salmo trutta L.). Heredity 82, 645. https://doi.org/10.1046/j.1365-2540.1999.00519.x
Pritchard, J.K., Stephens, M., Donnelly, P., 2000. Inference of population structure using multilocus
genotype data. Genetics 155, 945–959.
Pustovrh, G., Sušnik Bajec, S., Snoj, A., 2012. A set of SNPs for Salmo trutta and its application in
supplementary breeding programs. Aquaculture 370–371, 102–108.
https://doi.org/10.1016/j.aquaculture.2012.10.007
Rahel, F.J., 2007. Biogeographic barriers, connectivity and homogenization of freshwater faunas: it‘s a
small world after all. Freshw. Biol. 52, 696–710. https://doi.org/10.1111/j.1365-2427.2006.01708.x
Rahel, F.J., Nibbelink, N.P., 1999. Spatial patterns in relations among brown trout (Salmo trutta)
distribution, summer air temperature, and stream size in Rocky Mountain streams. Can. J. Fish.
Aquat. Sci. 56, 43–51. https://doi.org/10.1139/f99-210
Rengmark, A.H., Slettan, A., Skaala, Ø., Lie, Ø., Lingaas, F., 2006. Genetic variability in wild and farmed
Atlantic salmon (Salmo salar) strains estimated by SNP and microsatellites. Aquaculture 253, 229–
237. https://doi.org/10.1016/j.aquaculture.2005.09.022
Rosenberg, N.A., Li, L.M., Ward, R., Pritchard, J.K., 2003. Informativeness of genetic markers for
inference of ancestry. Am. J. Hum. Genet. 73, 1402–1422. https://doi.org/10.1086/380416
Rosenblum, E.B., Novembre, J., 2007. Ascertainment Bias in Spatially Structured Populations: A Case
Study in the Eastern Fence Lizard. J. Hered. 98, 331–336. https://doi.org/10.1093/jhered/esm031
Ryman, N., Palm, S., André, C., Carvalho, G.R., Dahlgren, T.G., Jorde, P.E., Laikre, L., Larsson, L.C.,
Palmé, A., Ruzzante, D.E., 2006. Power for detecting genetic divergence: differences between
statistical methods and marker loci: power for detecting divergence. Mol. Ecol. 15, 2031–2045.
https://doi.org/10.1111/j.1365-294X.2006.02839.x
Ryynanen, H.J., Tonteri, A., Vasemagi, A., Primmer, C.R., 2007. A Comparison of Biallelic Markers and
Microsatellites for the Estimation of Population and Conservation Genetic Parameters in Atlantic
Salmon (Salmo salar). J. Hered. 98, 692–704. https://doi.org/10.1093/jhered/esm093
Saint-Pé, K., Blanchet, S., Tissot, L., Poulet, N., Plasseraud, O., Loot, G., Veyssière, C., Prunier, J.G.,
2018. Genetic admixture between captive-bred and wild individuals affects patterns of dispersal in
a brown trout (Salmo trutta) population. Conserv. Genet. 19, 1269–1279.
https://doi.org/10.1007/s10592-018-1095-2
Sato, T., Harada, Y., 2008. Loss of genetic variation and effective population size of Kirikuchi charr:
implications for the management of small, isolated salmonid populations. Anim. Conserv. 11, 153–
159. https://doi.org/10.1111/j.1469-1795.2008.00165.x
Scheet, P., Stephens, M., 2008. Linkage disequilibrium-based quality control for large-scale genetic studies.
PLoS Genet. 4, e1000147. https://doi.org/10.1371/journal.pgen.1000147
Slettan, A., Olsaker, I., Lie, ø, 1995. Atlantic salmon, Salmo salar , microsatellites at the SSOSL25,
SSOSL85, SSOSL311, SSOSL417 loci. Anim. Genet. 26, 281–282. https://doi.org/10.1111/j.1365-
2052.1995.tb03262.x
Smith, S.M., Maughan, P.J., 2015. SNP genotyping using KASPar assays, in: Batley, J. (Ed.), Plant
Genotyping. Springer New York, New York, NY, pp. 243–256. https://doi.org/10.1007/978-1-
4939-1966-6_18
67
Splendiani, A., Ruggeri, P., Giovannotti, M., Pesaresi, S., Occhipinti, G., Fioravanti, T., Lorenzoni, M.,
Nisi Cerioni, P., Caputo Barucchi, V., 2016. Alien brown trout invasion of the Italian peninsula: the
role of geological, climate and anthropogenic factors. Biol. Invasions 18, 2029–2044.
https://doi.org/10.1007/s10530-016-1149-7
Stinchcombe, J.R., Hoekstra, H.E., 2008. Combining population genomics and quantitative genetics:
finding the genes underlying ecologically important traits. Heredity 100, 158–170.
https://doi.org/10.1038/sj.hdy.6800937
Suárez, J., Bautista, J.M., Almodóvar, A., Machordom, A., 2001. Evolution of the mitochondrial control
region in Palaearctic brown trout (Salmo trutta) populations: the biogeographical role of the Iberian
Peninsula. Heredity 87, 198–206. https://doi.org/10.1046/j.1365-2540.2001.00905.x
Sušnik Bajec, S., Pustovrh, G., Jesenšek, D., Snoj, A., 2015. Population genetic SNP analysis of marble and
brown trout in a hybridization zone of the Adriatic watershed in Slovenia. Biol. Conserv. 184, 239–
250. https://doi.org/10.1016/j.biocon.2015.01.033
Susnik, S., Schoffmann, J., Weiss, S., 2005. Genetic verification of native brown trout from the Persian
Gulf (Catak Cay River, Tigris basin). J. Fish Biol. 67, 879–884. https://doi.org/10.1111/j.0022-
1112.2005.00780.x
Swatdipong, A., Vasemägi, A., Niva, T., Koljonen, M.-L., Primmer, C.R., 2010. High level of population
genetic structuring in lake-run brown trout, Salmo trutta, of the Inari Basin, northern Finland. J.
Fish Biol. 77, 2048–2071. https://doi.org/10.1111/j.1095-8649.2010.02784.x
Tougard, C., Justy, F., Guinand, B., Douzery, E.J.P., Berrebi, P., 2018. Salmo macrostigma (Teleostei,
Salmonidae): Nothing more than a brown trout (S. trutta) lineage? J. Fish Biol. 93, 302–310.
https://doi.org/10.1111/jfb.13751
Vera, M., Cortey, M., Sanz, N., García-Marín, J.-L., 2010. Maintenance of an endemic lineage of brown
trout (Salmo trutta) within the Duero river basin. J. Zool. Syst. Evol. Res. 48, 181–187.
https://doi.org/10.1111/j.1439-0469.2009.00547.x
Vera, M., García-Marín, J.L., Martinez, P., Bouza, C., 2015. Phylogenetic diversity within the endemic
brown trout Duero lineage: implications for conservation and management. Mar. Freshw. Res. 66,
1066. https://doi.org/10.1071/MF14286
Vøllestad, L.A., Serbezov, D., Bass, A., Bernatchez, L., Olsen, E.M., Taugbøl, A., 2012. Small-scale
dispersal and population structure in stream-living brown trout (Salmo trutta) inferred by mark–
recapture, pedigree reconstruction, and population genetics. Can. J. Fish. Aquat. Sci. 69, 1513–
1524. https://doi.org/10.1139/f2012-073
Wang, S., Sha, Z., Sonstegard, T.S., Liu, H., Xu, P., Somridhivej, B., Peatman, E., Kucuktas, H., Liu, Z.,
2008. Quality assessment parameters for EST-derived SNPs from catfish. BMC Genomics 9, 450.
https://doi.org/10.1186/1471-2164-9-450
Waples, R.S., Hendry, A.P., 2008. Special Issue: Evolutionary perspectives on salmonid conservation and
management: Evolutionary perspectives on salmonids. Evol. Appl. 1, 183–188.
https://doi.org/10.1111/j.1752-4571.2008.00035.x
68
Supplementary material for Chapter I
Appendix S1.1 : RAD data set of 12204 sequences each containing one or two SNPs, with
minimum allele frequency of 5% or higher, no SNPs in the first or last 30 bp of the RAD, no
unsequenced nucleotides, and which can be positioned on the S. trutta linkage map
This file contains for each RAD, identification (Locus_ID ), chromosome on Atlantic Salmon
CHROMsalar), linkage group on the Brown trout linkage map (LG_Trutta),Position on the brown
trout linkage group (Pos_cM_Trutta), positions on the Atlantic salmon chromosome
(POSsalar_snp1, POSsalar_snp2), position of the SNP on the Rad tag (POStag_snp1,
POStag_snp2), recombination rate ( recombination_rate_cM.Mb), sequence of the tag (TAG),
sequencing direction (Orientation), and position of the rad (POS_RAD).
See file «12204RADsFinal.txt» in DOI: 10.6084/m9.figshare.8174708
Appendix S1.2: SNPs used for low density arrays (5 basins and Taurion test)
See file «SNPsLowDensityFinal245.txt» in DOI: 10.6084/m9.figshare.8174708
This file contains for each SNP, identification (SNP_ID), linkage group on the Brown trout
linkage map (LG_Trutta), position of the SNP on the Rad tag (POStag_snp), recombination rate
(recombination_rate_cM.Mb), sequence of the tag (TAG_sequence).
It also gives success at genotyping information:
- listSuccessTests162 : SNPs which have been genotyped (when TRUE) in the five river
basins tested for genotyping success.
- listTaurion182: SNPs which have been genotyped (when TRUE) in the five river basins
tested for genotyping success, and added SNPs which have been genotyped in the Taurion
basin (after removing the 30 which did not work in the five basins, then the 10 which did
not work in the Taurion).
69
Appendix S1.3: Isolation by distance patterns for each river basin, represented by plots of
pairwise Fst values against pairwise riparian distance between sites
70
Appendix S1.4: Pairwise Fst values (calculated with adegenet R package) between sites measured
with SNPs and with microsatellites
Pop BEA-Rau PON-Rau THA-Bar THA-Tcc THA-Usi THA-Vig VIG-Tex
SNP
s
PON-Rau 0.063 THA-Bar 0.0579 0.0173
THA-Tcc 0.0658 0.0175 0.0175 THA-Usi 0.0695 0.0215 0.0095 0.0193
THA-Vig 0.0762 0.0263 0.0126 0.0206 0.0158 VIG-Tex 0.0875 0.0337 0.0169 0.0238 0.012 0.0142
PIS-Sou 0.199907 0.153376 0.145969 0.153629 0.141601 0.138166 0.140239
Mic
rosa
telli
tes
PON-Rau 0.0389 THA-Bar 0.0312 0.0207
THA-Tcc 0.0349 0.0115 0.0102 THA-Usi 0.0397 0.0216 0.0127 0.0128
THA-Vig 0.0527 0.0329 0.019 0.0235 0.0121 VIG-Tex 0.0489 0.0291 0.0193 0.0222 0.0138 0.0141
PIS-Sou 0.0971 0.0626 0.0706 0.0677 0.0562 0.0664 0.0687
71
Appendix S1.5: Sample sizes and mean (± standard error) body length in mm of fish sampled in
each site and each basin for testing the 192 SNPs panel’s genotyping success.
AUDE DORON ONE ROYA SEUGE
Site N Mean body length (±SE)
Site N Mean body length (±SE)
Site N Mean body length (±SE)
Site N Mean body length (±SE)
Site N Mean body length (±SE)
Agu-Sou 27 157(±42) All-All 17 126(±55) Cou-Lar 30 111(±26) Ben-Cas 8 158(±33) Ber-Ben 11 114(±28)
Aig-Pou 30 139(±33) Boz-Car 12 173(±56) Lab-Cas 17 112(±23) Bie-Cas 30 157(±47) Bui-Pie 6 96(±27)
Aig-Sou 30 143(±38) Boz-Gib 15 127(±36) Nga-Jur 30 152(±40) Bie-Mai 30 172(±33) Cla-Tou 9 105(±55)
Art-Lau 30 138(±38) Boz-Vil 18 165(±41) Nga-Mar 30 135(±18) Bie-Min 30 171(±36) Lav-Riv 15 138(±51)
Aud-Car 30 166(±34) Cha-Chi 22 133(±40) Nga-Vga 30 127(±35) Cai-Gaf 30 170(±48) Lav-Suc 30 122(±31)
Aud-Far 30 165(±33) Chv-Ger 23 177(±69) Noo-Ast 30 135(±26) Lev-Bri 30 140(±27) Pon-Amo 30 127(±26)
Aud-Fou 30 135(±25) Pra-Fra 14 153(±62) Noo-Cas 30 144(±18) Lev-Ten 30 153(±30) Pon-Bom 30 133(±32)
Aud-Nen 30 123(±54) Ros-Mug 3 137(±72) Noo-Esp 30 129(±38) Mag-Cem 30 176(±32) Pon-Cha 30 148(±29)
Aud-Puy 30 121(±31)
Noo-Lac 30 109(±26) Ref-Ric 30 123(±23) Pon-Tis 30 129(±18)
Aud-Ser 30 156(±22)
Noo-Sav 30 111(±63) Roy-Chi 24 196(±24) Ser-Sau 6 143(±28)
Bai-Pas 14 153(±28)
Noo-Tre 30 157(±33) Roy-Dal 30 143(±27) Seu-Cch 30 110(±21)
Bru-Mij 30 127(±26)
Noo-Voo 30 138(±30) Roy-Ort 30 131(±18) Seu-Cou 30 134(±27)
Bru-Uss 30 151(±21)
Nou-Boo 28 173(±43) Roy-Pie 21 207(±31) Seu-Cro 30 165(±33)
Cam-Sau 14 118(±46)
Nou-Cir 30 134(±42) Roy-Sca 30 172(±47) Seu-Fag 30 126(±32)
Que-Mas 29 131(±36)
Nou-May 30 155(±34) Roy-Ten 30 141(±27) Seu-Rod 30 157(±38)
Que-Ria 30 158(±37)
Nou-Spo 30 140(±38) ROY-Fon 30 199(±65) Seu-Sau 30 138(±19)
Roq-Sau 9 154(±25)
One-Bag 30 147(±28) ROY-Evc 30 196(±59)
Roy-Bre 30 193(±63)
Roy-Amb 30 171(±52)
Roy-Gia 30 162(±47)
ROY-Vei 30 182(±40)
Total 453 143(±37) 124
150(±56) 495 136(±38) 413 167(±23) 377 134(±34)
72
Appendix S1.6: Map of the Taurion River showing sampling points (black dots). Sample sizes by
site, and mean (±SE) body length in mm) from the Taurion River in table. Map was generated by
authors on ArcGis and assembled using Inkscape.
Site N Mean (±SE) body length (mm)
BEA-Rau 30 111(±38)
PON-Rau 21 111(±37)
THA-Bar 28 221(±108)
THA-Tcc 30 144(±46)
THA-Usi 30 193(±40)
THA-Vig 26 167(±51)
VIG-Tex 25 151(±44)
The Taurion River is a Snow/rain fed stream from the Massif central (France). We sampled
between 21 and 30 individuals per site. 4 sites were located on the mainstream (THA sites), the 3
others on tributaries. These sampling sites are represented by black dots on the map.
73
Appendix S1.7: Genotyping microsatellites
Individual multilocus genotypes were obtained at a total of 13 microsatellite markers (BS131,
One9, SSosL311, SsoSL438, T3-13, Sfo1, Ssa064, Ssa103, Ssa417, Ssa-60NVH, Ssa85DU, Ssa-
TAP2a and SsoSL417; see Saint-Pé et al. [1] for details on these markers).
The 13 markers were assembled in 3 multiplexes allowing co-amplification of several markers in a
single Polymerase Chain Reaction (PCR). Genomic DNA was extracted from the fin clips using a
salt-extraction protocol [2]. The loci were amplified using the QIAGEN Multiplex PCR Kit
(Qiagen, Valencia, CA, USA). PCR reactions were carried out in a 10 µL final volume containing
5–20 ng of genomic DNA, 5 µL of 2xQIAGEN Multiplex PCR Master Mix, and locus-specific
optimized combination of primers. PCR amplifications were performed in a Mastercycler PCR
apparatus (Eppendorf, Hauppauge, NY, USA) under the following conditions: 15 min at 95°C
followed by 30 cycles of 1 min at 94°C, 1 min at 60°C and 1 min at 72°C and finally followed by
a 60 min elongation step at 72°C. Amplified fragments were then separated on an ABI PRISM
3730 automated capillary sequencer. Allelic sizes were finally scored using GENEMAPPER v.4.0
(Applied Biosystems, Foster City, CA, USA). We investigated for anomalies owed to genotyping
(e.g. large allele drop; null alleles) using MicrocheckerV 2.2 [3]. As in Saint-Pé et al. [1], we
tested for linkage disequilibrium and selection, which none of these 13 loci displayed of.
References Appendix S7
1. Saint-Pé K, Blanchet S, Tissot L, Poulet N, Plasseraud O, Loot G, et al. Genetic admixture between
captive-bred and wild individuals affects patterns of dispersal in a brown trout (Salmo trutta)
population. Conserv Genet. 2018;19:1269–79.
2. Aljanabi S. Universal and rapid salt-extraction of high quality genomic DNA for PCR- based techniques.
Nucleic Acids Res. 1997;25:4692–3.
3. Van Oosterhout C, Hutchinson WF, Wills DPM, Shipley P. micro-checker: software for identifying and
correcting genotyping errors in microsatellite data. Mol Ecol Notes. 2004;4:535–8.
74
Chapter II. Combining genetics, stable isotopes and
morphometrics improves assignments for movement
detection: a case study in Brown trout (Salmo trutta)
populations
Saint-Pé Keoni1*
, Julien Cucherousset2, Laurence Tissot
3, Nicolas Poulet
4, Simon Blanchet
1,2*, and
Jérôme G. Prunier1
1 Centre National de la Recherche Scientifique (CNRS), Université Paul Sabatier (UPS); Station
d‘Ecologie Théorique et Expérimentale, UMR 5321, F-09200 Moulis, France
2 CNRS, UPS, École Nationale de Formation Agronomique (ENFA) ; UMR 5174 EDB
(Laboratoire Évolution & Diversité Biologique), 118 route de Narbonne, F-31062 Toulouse cedex
4, France
3 EDF Recherche et Développement, Laboratoire National d‘Hydraulique et Environnement, 6
quai Watier, 78401 Chatou Cedex, France
4 Pôle écohydraulique AFB-IMT, allée du Pr Camille Soula, 31400 Toulouse France
Key words: Dispersal, brown trout, SNPs, isotopes, morphometrics, individual assignment.
75
Summary
Assessing landscape connectivity is of crucial importance for biodiversity dynamics, and
assignment methods are valuable approaches for this purpose. Assignment methods are generally
performed using a single data type such as genetic or isotopic markers. However, relying on a
single marker type for detecting movements can be sub optimal, for instance at small spatial scales
when populations are poorly differentiated. Here, we tested whether combining genetic, (stable)
isotopic and morphometric markers improves power and accuracy of assignments, and we
assessed the contribution of each data type to assignment rate. To do so, we developed an
approach based on a machine learning algorithm (random Forest) and a permutation procedure that
we applied to two brown trout (Salmo trutta) populations. We genotyped 830 brown trout using
SNP markers, quantified their stable isotope signatures (δ15N and δ13C values), and inferred they
morphometry using a geometrical approach. We found a significant increase in assignment power
when combining some data sets, in particular genetics and isotopes, as indicated by the detection
of nearly three times more residents as well as a several-fold reduced number of unassigned
individuals compared to these data types used alone. Then, we found that isotopes were the most
powerful data type to discriminate and assign individuals to their population of origin, while
morphometrics were the least powerful.
Our findings demonstrate the usefulness of combining marker types for inferring movements at
small spatial extents (within river basin), and that stable isotopes appeared particularly relevant for
identifying local scale movements. Finally, we provide an operational statistical framework to
generalize the use of multiple marker types to assess movements in natural populations.
76
Introduction
Landscape connectivity, the permeability of the landscape matrix to individual movements, has widely
been shown to be crucial for individuals to complete their life cycle (Dunning et al., 1992), ensuring
within- and between- population dynamics, and thus viability (Hanski, 1998; Yen, 1999; Hanski
and Gilpin, 1991; Wiens, 2001; Van Dyck and Baguette, 2005; Clobert, 2012; Wiens et al., 1997).
Understanding how habitat fragmentation, now considered as one of the first causes of biodiversity
loss (Wilson et al., 2016), affects population functioning through movements is therefore a growing
concern worldwide (Young et al. 1996; Fischer, and Lindenmayer 2007; Cote et al. 2017; Fahrig
and Merriam, 1994). Various methods and data types have been used to measure individual
movements (Cayuela et al., 2018, Kindlmann and Burel 2008). Directly measuring movement has
been done successfully using capture-recapture protocols, pit-tagging and telemetry for instance
(Cooke et al., 2013). However, it is a difficult, expensive and time consuming task that rarely
allows for large sample sizes (Koenig et al., 1996; Birnie-Gauvin et al., 2018).
More recently developed, genetic-based methods allow indirect estimations of movements, under
the hypothesis that population genetic structure reflects the impact of natural and anthropogenic
features on patterns of landscape functional connectivity (Davey et al., 2011; Forester et al., 2018;
Manel et al., 2003; Holderegger et al., 2006; Broquet and Petit, 2009; Baguette et al., 2013).
Genetic data have traditionally been used to assess patterns of effective dispersal, that is, inter-
population movements followed by reproduction events, using methods such as edge detection
methods, clustering algorithms, parentage analyses, Fst based estimates, and many others (see
Cauyela et al. 2018 for a recent review). However, these methods reflect rather long-term dispersal
events, with a deferred response regarding ‗real-time‘ changes in movement patterns (Landguth et
al. 2010, Whitlock and McCauley, 1999).
Individual-based genetic assignment methods, based on the use of genotypic data to discriminate
among resident and immigrant individuals in local populations and determine the population of
origin of immigrants, appear as an interesting alternative for rapid estimation of ―real-time‖
movements. Indeed, they do not rely on any theoretical assumptions as other genetic tools and
allow detecting different kinds of movements, such as non-effective dispersal events, but also
migration events and/or foraging movement (Paetkau et al., 2004; Raeymaekers et al., 2008; Saint-
77
Pé et al., 2018). Genetic assignment methods are highly accurate for studying movement
particularly in cases where dispersal is unlikely, thus between distant and/or well differentiated
populations (Berry et al., 2004; Manel et al., 2002; Paetkau et al., 1995). However, they were also
shown to be poorly accurate between less differentiated populations, for instance at fine spatial
scale (Anderson et al., 2008; Berry et al., 2004; Manel et al., 2005).
Although assignment procedures have been initially developed for genetic markers, these
approaches can use other data types, and some studies have recently take advantage of this (Begg
and Waldman, 1999). Data types that have been successfully used for individual assignments are
various, and include morphological traits (Bagherian and Rahmani, 2007; Caccamise et al., 2000),
parasites (Timi et al., 2005), fatty acid composition (Czesny et al., 2000), and elemental or isotopic
composition contains in solid structures (fish otoliths: Martin et al., 2013; bird feathers: Hobson,
1999; invertebrate shells: Becker et al., 2007). They generally provide complementary and
alternative information to genetic markers, mainly because they vary at different spatial and
temporal scales than genetic markers (Cook et al., 2007). Moreover, combining different maker
types may provide better discrimination among individuals and populations than genetics alone,
and thus higher assignment accuracy and power (Clegg et al., 2003; Gómez-Díaz and González-
Solís, 2007; Kelly et al., 2005; Smith and Campana, 2010). An interesting example of how
coupling methods can better capture movement events concern Atlantic salmon (Salmo salar) in
which descendants of hatchery-born Salmon were shown to be born in the wild using genetics and
geochemistry of otoliths, whereas they would have been identified as hatchery-raised juveniles if
considering only genetic data (Perrier et al. 2011).
Recently, new methods of supervised machine learning (Kotsiantis, 2007) were developed, and
appeared very efficient for robustly discriminating among populations and assigning individuals to
their populations or origin with the use of different types of data. Supervised learning aims at
predicting the value of either a categorical or a continuous variable through the use of a training
set of labeled data whose true outcome response is known, to ultimately train a predictive
algorithm. Among these methods, Random forests in particular are highly interesting for this
purpose (Chen et al., 2018; Cutler et al., 2007). They consist in ―bagged decision tree models‖, i.e.
predicting a value or a class by averaging the value obtained by many different uncorrelated
decision trees (Breiman, 2001). Moreover, these methods present high versatility and low bias, do
78
not suffer from the numerous requirements and assumptions of classical (genetic) assignment
methods, and have already been used successfully for assignment purposes (Guinand, 2002;
Schrider and Kern, 2018; Sheehan and Song, 2016).
Rivers are particularly concerned with fragmentation, and conservation of these ecosystems is of
growing concern (Dudgeon et al., 2006; Birnie-Gauvin et al., 2017; Reid et al., 2018). There are
more than 45 000 large dams in the world (Duflo and Pande 2007), and in France only for
instance, the estimated number of riverine obstacles is above 80 000 (ROE: référentiel des
obstacles à l'écoulement). Their effects on riverine fish notably are of high concern (Calò et al.,
2013; Heggenes et al., 2006; Junker et al., 2012), and have thus been widely studied using
assignment methods (Barnett-Johnson et al., 2010; Morrissey and Ferguson, 2011; Saint-Pé et al.,
2018). However, studies combining data types are scarce, and in most cases, occur at very large
scales and/or concern highly differentiated populations within contrasted environments. For
instance, isotopes are often used as a tool to measure movement at very large scales, between
freshwater and saltwater movements (Acolas et al., 2008; McCarthy and Waldron, 2000), or
movements between anthropized and natural environments, the former being characterized with
enhanced δ15
N (Hansson et al., 1997). To our knowledge, no studies have combined different data
types to assess fish movements at a within-basin fine-scale resolution.
Here, we used a headwater stream fish, the brown trout (Salmo trutta), in two independent river
catchments (i) to assess the benefits of combining genetics, carbon and nitrogen stable isotopes,
and morphometrics in assigning individuals to their population of origin and (ii) to identify which
of these markers is the most powerful for assignment. We used a random forest classification
algorithm that we combined to a robust permutation procedure to determine individual
probabilities of originating from each sampled site. We apply this approach to every possible
combination of data (i.e. genetics only, isotopes only, morphometrics only, genetics + isotopes,
genetics + morphometrics, isotopes + morphometrics, and genetics + isotopes + morphometrics) to
assess which data type(s) can be of use at this spatial extent, and how combining datasets can
improve assignment power. At this relatively small spatial scale, we expected between-population
discrimination to be low, regardless the data type used. However, different data types are supposed
to be uniquely distributed within landscapes for different reasons: genetic drift and selection for
genetic markers (Hartl and Clark, 2007), geology and trophic interactions for isotopes (Peterson
79
and Fry, 1987), phenotypic responses to environmental constraints for morphometrics, either
through selection or plasticity (Cunico and Agostinho, 2006). We thus expected that by combining
data types, populations should be better discriminated than with only one data type used. Higher
discrimination between populations should provide better assignment accuracy, that is, higher
assignment probabilities for each individual, which in turn translates into a reduced number of
unassigned individuals, and into the detection of more ―residents‖ (Manel et al., 2002; Paetkau et
al., 1995). Combining data types should also imply higher statistical power, with a reduced type I
and type II error rates (i.e. erroneously classifying a resident as migrant or a migrant as resident,
respectively; Maxwell et al., 2008). Following these ideas, we expected that combining all datasets
(genetics + isotopes + morphometrics) should provide the highest discrimination between brown
trout populations, and therefore the highest assignment accuracy and statistical power. Finally, we
discuss the idea that data types might convey unique biological information about the spatio-
temporal dynamics of different kinds of movements (foraging, migration, natal dispersal), further
highlighting the potential benefits of considering different data types in supervised assignment
methods.
80
Methods
Field sampling
We sampled two independent French river basins (Figure 1) in which the fish community is
dominated by brown trout (Salmo trutta): the Aude River (Mediterranean outlet) and the Seuge
River (Atlantic outlet). Brown trout were sampled using a single-pass electric-fishing protocol
from 33 sites in total, with an aim of 30 individuals per site. We focused on small individuals in
order to only sample one age class (1 year-old), based on unpublished length-frequency
distributions (Tissot et al., 2016-2018) as we did not use scale analysis. In total, we captured 830
individuals (453 from the Aude River, 377 from the Seuge River), with sample sizes ranging from
6 to 30 per site. Each individual was measured (total length in mm; Table 1) and a pelvic fin clip
was taken for genetic and isotopic analyses. All individuals were released alive to their original
sampling site.
Figure 2.1: Location of the two studied systems in France and locations of sampled sites within each basin.
The Seuge River is part of the Atlantic lineage, whereas the Aude River is part of the Mediterranean
lineage. Thirty three sampling sites have been selected within each river basin (black dots on river
networks).
81
Table 2.1 Number of individuals sampled and their average (±Standard error) body length for each site (in
mm).
Genotyping
Individual multilocus genotypes were obtained for 162 SNPs markers (Saint-Pé et al., 2018), using
KASPAR technology performed by LGC Genomics (Smith and Maughan, 2015). All polymorphic
markers were kept regardless of selection and linkage disequilibrium, as these processes may
increase assignment power and may thus be exploited for investigating population structure
(Morin et al., 2009; Nielsen et al., 2009; Waples and Gaggiotti, 2006). For further analyses, we
discarded individuals for which more than one third of SNPs were not genotyped, leaving us with
824 individuals (449 in the Aude River, 375 in the Seuge River).
As stocking is a common practice in brown trout and has been shown to affect dispersal and
spatial patterns of genetic diversity in the river basins studied here (Saint-Pé et al., 2018), we
decided to filter out all individuals showing more than 30% of captive-bred ancestry in order to
only keep supposedly native individuals. Based on genetics, individual level of hatchery ancestry
was determined using STRUCTURE (Pritchard et al., 2000) as in Saint-Pé et al. (2018, 2019).
AUDE SEUGE
Site N Mean Body length (±SE) Site N Mean Body length (±SE)
Agu-Sou 27 157(±42) Ber-Ben 11 114(±28) Aig-Pou 30 139(±33) Bui-Pie 6 96(±27) Aig-Sou 30 143(±38) Cla-Tou 9 105(±55) Art-Lau 30 138(±38) Lav-Riv 15 138(±51) Aud-Car 30 166(±34) Lav-Suc 30 122(±31) Aud-Far 30 165(±33) Pon-Amo 30 127(±26) Aud-Fou 30 135(±25) Pon-Bom 30 133(±32) Aud-Nen 30 123(±54) Pon-Cha 30 148(±29) Aud-Puy 30 121(±31) Pon-Tis 30 129(±18) Aud-Ser 30 156(±22) Ser-Sau 6 143(±28) Bai-Pas 14 153(±28) Seu-Cch 30 110(±21) Bru-Mij 30 127(±26) Seu-Cou 30 134(±27) Bru-Uss 30 151(±21) Seu-Cro 30 165(±33) Cam-Sau 14 118(±46) Seu-Fag 30 126(±32) Que-Mas 29 131(±36) Seu-Rod 30 157(±38) Que-Ria 30 158(±37) Seu-Sau 30 138(±19) Roq-Sau 9 154(±25)
Total 453 143(±37) 377 134(±34)
82
This only concerned the Aude basin (48 individuals), as the Seuge basin showed only very slight
traces of hatchery ancestry (all individuals assigned at less than 0.3 to the captive-bred cluster).
The native genetic dataset comprised 776 individuals (401 for the Aude River, 375 for the Seuge
River), and only these individuals were kept for subsequent analyses.
Carbon and Nitrogen stable isotopes
For all 776 native individuals, fin samples were dried at 60 °C for 72 hours, and then loaded into
1.5mL Eppendorf tubes, labeled and parafilmed. These fins were analyzed at Cornell University
(COIL, Ithaca NY), using an elemental analyzer plumbed into a Finnigan MAT Delta Plus isotope
ratio mass spectrometer to measure samples for δ13C
and δ15
N. Isotope analysis failed for three of
the individuals, which were therefore discarded from all analyses. The final isotope dataset
comprised 773 individuals (399 for the Aude River, 374 for the Seuge River).
Morphometrics
While on the sampling sites, each individual was placed (after having been anaesthetized) on a
polystyrene board with a ruler and a color sampler, and was photographed with a digital camera,
always at the same magnification (X 2.8). Using the TPS suite (tpsDig 2.10 and tpsUtil 1.38;
Rohlf, 2015), 12 morphological landmarks (Details in Appendix S1) were digitized from each fish.
We then used the GeoMorph R package (Adams and Otárola-Castillo, 2013) to calculate the
position of each individual along relative warps, which consist in major axes of body shape
variation after Principal Components Analyses (Park et al., 2013). We kept the first six relative
warps (75.7% of explained variance) in order to explain most of the morphometric variation.
However, as the first relative warp represented a deformation that likely result from a bias in fish
positioning on the board (29.8% of explained variance), we discarded it from the morphometrics
dataset.
The final dataset therefore comprised 773 individuals with all completed data types.
Assignment method
First, and for each combination of datasets (i.e. G = genetics only, I = isotopes only, M =
morphometrics only, G+I = genetics + isotopes, G+M = genetics + morphometrics, I+M =
83
isotopes + morphometrics, and G+I+M = genetics + isotopes + morphometrics), we calculated
individual assignment probability to each site using the random forest classification method (R-
package randomForest; Liaw and Wiener, 2002) with the following formula: ―site ~ data
combination‖. The number of trees as well as the number of splits of the variables (―ntree‖ and
―mtry‖ parameters, respectively) were chosen to provide the lowest and more stable out-of-bag
error rate (OOB; classification to sites different from capture site).
When markers are poorly discriminant (e.g. between populations are poorly differentiated), Type I
(erroneously classifying a resident individual as a migrant) and Type II (erroneously classifying a
migrant individual as a resident) errors can be high, and can be difficult to identify. In order to set
a conservative threshold in assignment probabilities and limit these errors, we developed a
permutation-based procedure. The aim of this procedure was to calculate the theoretical maximum
probability of assignment to a site under the assumption that sites were totally undifferentiated, so
as to increase the overall statistical power of the assignment procedure and limit type I and II
errors (at the cost of more unassigned individuals that cannot confidently be assigned to a
population). To do so, we performed 1000 random permutations of individuals across capture sites
and ran RF classification on all the 1000 permuted datasets, with the same number of trees and
dataset splits as in the first step, to compute the 95th
quantile of maximum possible assignment
values over the 1000 runs.
Following these classification and permutation procedures, individuals were considered as
―resident‖ if the observed probability of assignment was the highest for their capture site (which is
conservative and the most parcimonious). Individuals were considered as ―migrant‖ if the
observed probability of assignment was highest for another site than its site of capture, and only if
this probability was higher than the theoretical maximum assignment probability expected if
populations are undifferentiated and as computed from permutations. Individuals assigned to a site
different from their capture site, but with a probability lower than the theoretical assignment
probability were considered as ―unknown‖ (as for these individuals type I errors are expected to be
high). This approach was performed for each basin separately and for each combination of
variables (G, I, M, G+I, G+M, I+M and G+I+M).
84
Contribution of combining data types
First, we used the mean decrease in Gini coefficient (Breimann et al. 2001) to calculate the
contribution of each variable type (genetics, isotopes and morphometrics) to the randomForest
model using the G+I+M combination. The mean decrease in Gini coefficient represents the mean
decrease in accuracy of the model due to the exclusion (or permutation) of a variable, which is
determined during the OOB error calculation: the more the accuracy decreases due to the
exclusion of a variable (or group of variables in our case), the more important that variable or
group of variables is for classification of the data and individual assignments.
Then, we assessed the assignment accuracy of each data combination across river basins. If
combining different data types yields higher inter-population discrimination, the assumption that
we considered as the most parsimonious is that the inferred number of residents should be higher.
We thus considered individual assignment probability to the capture site as a proxy for assignment
accuracy. Differences in accuracy between combinations of variables were assessed using a linear
mixed-effect model (lme function from the nlme R-package), with the individual assignment
probability to the capture site as the dependent variable, the data combination as a fixed effect, and
sites nested within river basins as random effects. As we expected that combining all datasets
(G+I+M) should provide the highest assignment accuracy, we used the relevel R function to set
this combination as the reference factor in our model. Therefore, the model provides the gain/loss
in accuracy of all possible data combinations compared to the full combination of genetic, isotopic
and morphometric data.
To assess the differences in assignment accuracy between basins, we investigated the conditional
modes of the basin random effects using the ranef function from the nlme R-package. Conditional
modes measure the difference between the average predicted response in each basin and the
response predicted for a particular individual.
Individual decisions: migrant, resident, unknown
Finally, for each basin, we counted individuals considered as migrants, residents and unknown
after the permutation threshold was applied for each data type and combination (G, I, M, G+I,
G+M, I+M and G+I+M). The first objective was here to estimate whether combining datasets
85
would indeed increase between site discrimination, and thus accuracy (under the hypothesis that
increasing accuracy results in maximizing the number of residents as discussed above). The
second objective was to assess the differences in statistical power between data types and
combinations, and therefore to determine which combinations enabled to minimize the number of
unassigned (―unknown‖) individuals.
Results
Data types and combinations comparison
Relative contributions of each data type to the model combining them all (G+I+M) are shown in
Figure 2. The contribution index was much higher for isotopes than for the two other markers,
especially in the Aude River. Surprisingly, genetics contributed little to this model, less than
morphometrics, although assignment accuracy was slightly higher for genetics, as indicated below.
Figure 2.2: Mean decrease in Gini, i.e. the contribution of each variable type to the randomForest model
when all variables are used in combination to explain individual assignment probability to each sampling
site.
Assignment accuracy results for each marker combination and for each river basin are shown in
Figure 3. Overall, the combination of genetics and isotopes (G+I) showed the highest assignment
accuracy (although Isotopes alone were slightly more accurate in the Aude River). The
combination of genetics, isotopes and morphometrics (G+I+M) was not significantly less accurate
86
(p-value = 0.91, Table 2A), although including morphometric markers did not increase accuracy.
Conversely, all other combinations (and marker types considered alone) were significantly less
accurate (p-value < 0.001, Table 2A), in particular morphometrics (M, -0.52 in accuracy compared
to the reference combination G+I+M), genetics (G, -0.41 in accuracy compared to the reference
combination) and their combination (G+M, -0.43 in accuracy compared to the reference
combination).
Figure 2.3: Assignment accuracy boxplots for all data types combinations and each river basin.
87
A Value Std.Error DF t-value p-value
Intercept (G+I+M) 0.645 0.050 5399.000 12.849 0.000
G -0.408 0.012 5399.000 -35.394 0.000
G+I 0.001 0.012 5399.000 0.118 0.906
G+M -0.433 0.012 5399.000 -37.568 0.000
I -0.045 0.012 5399.000 -3.896 0.000
I+M -0.160 0.012 5399.000 -13.899 0.000
M -0.523 0.012 5399.000 -45.351 0.000
Random effects (Intercept) Residual
~1|BV StdDev 0.065
~1|Site/BV StdDev 0.105 0.227
B Ranef, level=BV
Aude 0.042
Seuge - 0.042
Table 2.2: Summary of the linear mixed-effect model explaining assignment accuracy with data
combination used as a fixed effect, and sites nested within river basins used as random effects. (A)
Gain/loss in assignment accuracy for all combinations compared to the reference combination (G+I+M);
(B) Difference in assignment accuracy between basins across all data combinations.
When comparing averaged assignment accuracy between basins, we found that assignment
accuracy was lower in the Seuge River. Indeed, assignment accuracy was overall 0.084 higher in
the Aude River than in the Seuge River (Table 2B, Figure 3), however, the uncertainty (standard
deviation = 0.065) on the basin effect was higher than the deviation to the intercept due to random
effect of the basin.
Counts of individuals detected as migrants, residents or unassigned
The percentage of migrants, residents and unknown individuals for each combination of markers
used are shown in figure 4. As expected, data combinations with low assignment accuracy led to
the highest numbers of both migrants and unknown individuals (Figure 4: G, M and G+M for
instance). Interestingly, we found that the two combinations of data types that maximized
assignment accuracy (i.e., that maximized the number of individual actually assigned to their
capture site; G+I and G+I+M) did not classify any individual as unknown except in the Seuge for
the combination G+I+M, despite a dedicated permutation procedure designed to minimize Type 1
88
error rate by setting conservative theoretical thresholds. These two combinations thus showed both
high assignment accuracy and statistical power.
Figure 2.4: Number of migrants, residents and unclassified individuals for all data types combinations and
each river basin.
Discussion
Increasing statistical power in assignment procedures
We used three marker types that could potentially be of interest to detect movements at the spatial
extents we considered, i.e. headwater river basins of a few tens of kilometers across. First,
genetics, which have already been used very widely for this purpose, rely on inter-population
differentiation to enable sufficient discrimination among populations for detecting movements. As
the brown trout exhibits strong population structure at relatively small spatial extents (Carlsson et
al., 1999; Jensen et al., 2006), we expected that these markers would be particularly powerful in
our case studies. Second, isotopes are also widely used to study movements, however, as they are
mainly used in contrasted environments, we expected that their discriminant ability and thus their
89
power would be lower than genetics in the river basins studied. Finally, we also expected that
morphometrics should be poorly discriminant mainly because differential selection pressures
should be low at such reduced spatial scales.
Methodologically, the framework we used is similar to that developed in Chen et al. (2018), where
assignment accuracy is compared using different data types alone and in combination. However,
conversely to Chen et al. (2018), we stuck to random forest classifications to set conservative
thresholds in assignment probabilities using a permutation approach we developed, instead of
using existing resampling methods. Similarly to these methods, this enabled to diminish Type I
and II assignment error rates, particularly in the case of lowly discriminant data. Individuals that
do not meet theoretical thresholds are discarded (classified as unknown), which implies a gain in
assignment power yet with an increase in the number of unassigned individuals as a counterpart.
Morphometrics exhibited the lowest accuracy and power for assignment. This result is surprising,
as most morphological traits have a polygenic basis and are heavily influenced by environmental
factors through plasticity (Allendorf and Leary, 1988; Keeley et al., 2007). Diet and vegetation
cover (Svanbäck and Eklöv, 2006) for instance were shown to strongly act on morphology in fish
even at small spatial scales, and in particular for salmonids, water velocity (Pakkasmaa and
Piironen, 2000). At the spatial scales considered, selection on morphological features was
expected to be relatively high (Stelkens et al., 2012). We thus suggest that differences in
morphology at these scales may be very context dependent, thus in our study systems, the
environmental variables acting on morphometrics probably vary insufficiently to differentially
select morphotypes. Moreover, even if factors such as competition for resource can select for
different morphotypes (Ward et al., 2006), these morphotypes may also be living in sympatry
(Moore and Bronte, 2001) in our study systems, thus being unusable for discrimination between
populations.
Genetics were not as accurate and powerful as expected for assignment, especially for the Aude
River, in which almost half of individuals were not assigned, although less SNPs were available
for the Seuge River. This result is in accordance with the expectation that genetic assignment is
often more efficient when populations are more genetically differentiated (Berry et al., 2004), as
suggested by higher levels of genetic uniqueness measured in the Seuge River than in the Aude
90
River (average 0.074 in the Seuge, 0.042 in the Aude; Appendix S2). Each river basin being
unique, discrimination between populations, and thus of assignment accuracy and power will
strongly depend on the history of the river considered in terms of colonization history,
demographic history (bottlenecks for instance), selection patterns and many other factors such as
the panel of selected markers for instance (Banks et al., 2003). We thus suggest that in cases in
which genetic differentiation is low at the ―sampling site‖ level, assignment and migrant detection
should be more accurate at the ―cluster‖ level (Berry et al., 2004). However, in terms of
management and straightforwardness for instance, working at this abstracted level could be
problematic for both sampling designs and interpretation. Therefore in these low genetic
differentiation cases, other markers can here be highly valuable, and in our results, isotopes
appeared an excellent alternative, particularly in the Aude basin.
Isotopes exhibited the highest assignment accuracy and power. This result was surprising, as we
were expecting low inter-population variation given the reduced spatial extent of the studied areas,
which is expected to translate into low inter-population differences in isotopic signatures of
resources (Ben-David and Flaherty, 2012; Gray et al., 2004; Kennedy et al., 2005). Moreover,
resource competition could also have been expected to increase intra-population divergence as a
result of specialization at the individual level (Ward et al., 2006). However this was not observed
in our results, suggesting that contrary to the variables which act on morphometrics, those acting
on isotopic signatures (i.e. geochemical processes) are very contrasted between sites at this spatial
extent. This hypothesis was also supported by a similar pattern in periphyton and in invertebrate
prey, whose isotopic signatures also exhibited high inter-population and low intra-population
variation (Appendix S3). Isotopes therefore appeared as highly discriminant, thus accurate and
powerful at the extent of our study system. This is also of great interest because being highly
accurate and powerful at the ―sampling site‖ scale, (whereas genetics could bore useful at the
―cluster‖ scale), it can have direct applications, in particular for managers (e.g. for targeting a
specific location and identifying specific landscape features which could be constraining
movements).
Although in some cases, one data type can be more discriminating and powerful than another, one
of our most valuable results is that combining datasets was the best way to gain in assignment
efficiency. For instance, if genetics alone were more powerful in the Seuge than in the Aude River,
91
it was the opposite in the case of isotopes, but by combining genetics and isotopes, accuracy and
power were gained in both rivers, although they were still higher in the Aude River, mostly
because isotopes were extremely discriminating in this basin. Moreover, we found that the
combination of genetics and isotopes was the most powerful, as it was the only combination for
which there were no unassigned individuals. We thus argue that combining data types in
assignment protocols can highly improve the detection of movements occurring between
populations, as suggested in Chen et al. (2018), thus providing valuable information in terms of
functional connectivity (Holderegger et al., 2006; Broquet and Petit, 2009; Baguette et al., 2013).
However, some cases have shown that this is not always true, for instance, Smith and Campana,
(2010) found that combining genetics and otoliths isotopes decreased assignment success. As
suggested in Chen et al. (2018), our study also highlights that methods for integrating different
data types are still to be improved for bettering assignment.
Considerations when combining different data types
The advantage of combining different data types appeared as a very valuable method to increase
assignment statistical performance. However, different data types may be associated with different
spatio-temporal resolutions and may thus capture different biological processes at different scales
(Hall and Beissinger, 2017). For instance, and contrary to the genetic signature of an individual,
the isotopic signature of an individual may change over time depending on diet and isotopic
turnover rate (Hobson and Wassenaar 2008): assignment procedures based on isotopes will thus
only allow detecting short-term movements, depending on the capture vs. movement timing.
Therefore, and this has rarely been discussed in assignment methods studies, combining data types
such as genetics and isotopes could blur the biological interpretation of inferred movement
patterns. We thus suggest that comparing data types in addition to combining them might provide
unprecedented biological insights as to the studied movement patterns, although assignment
procedures would then lose in accuracy and statistical power. For instance, an individual captured
soon after emigrating and before isotopic turnover would be correctly detected as ―migrant‖ with
both genetics and isotopes, whereas an individual captured long after dispersal would be assigned
to its population of origin and thus detected as ―migrant‖ using genetics, but to the population
where it was recently foraging and thus detected as ―resident‖ when using isotopes. On the
opposite, for instance in the case of seasonal migration events, an individual captured right after it
92
returned to its site of origin would be detected as ―migrant‖ using isotopes, but detected as
―resident‖ using genetics.
In our study for instance, the higher proportion of individuals detected as residents when using
isotopes than when using genetics (suggesting higher assignment accuracy) could hide the fact that
a high proportion of inferred residents are actually longer-term migrants, with a short-term
―resident‖ isotopic signatures but a long-term ‖migrant‖ genetic signature. These hypotheses
might of course require further investigation, but similar considerations have been discussed for
morphometrics of developing larvae (therefore with a short ―turnover time‖), where variation
among populations differed through time, demonstrating the importance of evaluating divergence
over multiple time periods and at a time relevant to the processes being studied (Zellmer, 2018).
Nevertheless, we believe that combining different data types may be the most valuable tool when
the aim of the study is only to assess landscape functional connectivity regardless of the type of
movement and underlying biological mechanisms, as we demonstrated that this procedure
provides greater discrimination, and thus accuracy and statistical power.
Conclusions
Our results showed that different data types can be used separately and combined to increase
accuracy in individual assignment probabilities. In particular, stable isotopes of Carbon and
Nitrogen were very accurate tools at a short spatial extent, whereas genetic data did not show such
a discriminatory power among populations. Cheaper and less time consuming than the use of
genetics and morphometrics, our study suggests that isotopes thus have great potential to study
movement in stream fish at a small spatial scale. Combining different data types may yield very
high assignment accuracy and statistical power, providing valuable insight in terms of landscape
functional connectivity. Concomitantly, comparing assignment probabilities based on marker
types with different spatial and temporal resolutions may offer unprecedented opportunities to
characterize movement patterns, although future studies should now be conducted to further assess
the possible benefits of such comparisons.
93
References
Acolas, M.-L., Roussel, J.-M., Baglinière, J.-L., 2008. Linking migratory patterns and diet to reproductive
traits in female brown trout (Salmo trutta L.) by means of stable isotope analysis on ova. Ecology
of Freshwater Fish 17, 382–393. https://doi.org/10.1111/j.1600-0633.2007.00290.x
Adams, D.C., Otárola-Castillo, E., 2013. geomorph: an R package for the collection and analysis of
geometric morphometric shape data. Methods in Ecology and Evolution 4, 393–399.
https://doi.org/10.1111/2041-210X.12035
Allendorf, F.W., Leary, R.F., 1988. Conservation and distribution of genetic variation in a polytypic
species, the Cutthroat trout. Conservation Biology 2, 170–184. https://doi.org/10.1111/j.1523-
1739.1988.tb00168.x
Anderson, E.C., Waples, R.S., Kalinowski, S.T., 2008. An improved method for predicting the accuracy of
genetic stock identification. Canadian Journal of Fisheries and Aquatic Sciences 65, 1475–1486.
https://doi.org/10.1139/F08-049
Bagherian, A., Rahmani, H., 2007. Morphological differentiation between two populations of the Shemaya,
Chalcalburnus chalcoides : a geometrical morphometric approach. Zoology in the Middle East 40,
53–62. https://doi.org/10.1080/09397140.2007.10638204
Banks, M.A., Eichert, W., Olsen, J.B., 2003. Which genetic loci have greater population assignment
power? Bioinformatics 19, 1436–1438. https://doi.org/10.1093/bioinformatics/btg172
Barnett-Johnson, R., Teel, D.J., Casillas, E., 2010. Genetic and otolith isotopic markers identify salmon
populations in the Columbia River at broad and fine geographic scales. Environmental Biology of
Fishes 89, 533–546. https://doi.org/10.1007/s10641-010-9662-5
Becker, B.J., Levin, L.A., Fodrie, F.J., McMillan, P.A., 2007. Complex larval connectivity patterns among
marine invertebrate populations. Proceedings of the National Academy of Sciences 104, 3267–
3272. https://doi.org/10.1073/pnas.0611651104
Begg, G.A., Waldman, J.R., 1999. An holistic approach to fish stock identification. Fisheries Research 43,
35–44. https://doi.org/10.1016/S0165-7836(99)00065-X
Ben-David, M., Flaherty, E.A., 2012. Stable isotopes in mammalian research: a beginner‘s guide. Journal of
Mammalogy 93, 312–328. https://doi.org/10.1644/11-MAMM-S-166.1
Berry, O., Tocher, M.D., Sarre, S.D., 2004. Can assignment tests measure dispersal? Molecular Ecology 13,
551–561. https://doi.org/10.1046/j.1365-294X.2004.2081.x
Breiman, L., 2001. [No title found]. Machine Learning 45, 5–32. https://doi.org/10.1023/A:1010933404324
Caccamise, D.F., Reed, L.M., Castelli, P.M., Wainright, S., Nichols, T.C., 2000. Distinguishing migratory
and resident Canada geese using stable isotope analysis. The Journal of Wildlife Management 64,
1084. https://doi.org/10.2307/3803219
Calò, A., Félix-Hackradt, F.C., Garcia, J., Hackradt, C.W., Rocklin, D., Treviño Otón, J., Charton, J.A.G.,
2013. A review of methods to assess connectivity and dispersal between fish populations in the
Mediterranean Sea. Advances in Oceanography and Limnology 4, 150–175.
https://doi.org/10.1080/19475721.2013.840680
Carlsson, J., Olsen, K.H., Nilsson, J., Overli, O., Stabell, O.B., 1999. Microsatellites reveal fine-scale
genetic structure in stream-living brown trout. Journal of Fish Biology 55, 1290–1303.
https://doi.org/10.1111/j.1095-8649.1999.tb02076.x
Cayuela, H., Rougemont, Q., Prunier, J.G., Moore, J.S., Clobert, J., Besnard, A. and Bernatchez, L., 2018.
Demographic and genetic approaches to study dispersal in wild animal populations: A
methodological review. Molecular ecology, 27(20), pp.3976-4010.
Chen, K.-Y., Marschall, E.A., Sovic, M.G., Fries, A.C., Gibbs, H.L., Ludsin, S.A., 2018. assignPOP: An R
package for population assignment using genetic, non-genetic, or integrated data in a machine-
learning framework. Methods in Ecology and Evolution 9, 439–446. https://doi.org/10.1111/2041-
210X.12897
94
Clegg, S.M., Kelly, J.F., Kimura, M., Smith, T.B., 2003. Combining genetic markers and stable isotopes to
reveal population connectivity and migration patterns in a Neotropical migrant, Wilson‘s warbler
(Wilsonia pusilla). Molecular Ecology 12, 819–830. https://doi.org/10.1046/j.1365-
294X.2003.01757.x
Cook, B.D., Bunn, S.E., Hughes, J.M., 2007. Molecular genetic and stable isotope signatures reveal
complementary patterns of population connectivity in the regionally vulnerable southern pygmy
perch (Nannoperca australis). Biological Conservation 138, 60–72.
https://doi.org/10.1016/j.biocon.2007.04.002
Cooke, S.J., Midwood, J.D., Thiem, J.D., Klimley, P., Lucas, M.C., Thorstad, E.B., Eiler, J., Holbrook, C.,
Ebner, B.C., 2013. Tracking animals in freshwater with electronic tags: past, present and future.
Animal Biotelemetry 1, 5. https://doi.org/10.1186/2050-3385-1-5
Cunico, A.M., Agostinho, A.A., 2006. Morphological patterns of fish and their relationships with reservoirs
hydrodynamics. Brazilian Archives of Biology and Technology 49, 125–134.
https://doi.org/10.1590/S1516-89132006000100015
Cutler, D.R., Edwards, T.C., Beard, K.H., Cutler, A., Hess, K.T., Gibson, J., Lawler, J.J., 2007. Random
forests for classification in ecology. Ecology 88, 2783–2792. https://doi.org/10.1890/07-0539.1
Czesny, S., Dabrowski, K., Christensen, J.E., Van Eenennaam, J., Doroshov, S., 2000. Discrimination of
wild and domestic origin of sturgeon ova based on lipids and fatty acid analysis. Aquaculture 189,
145–153. https://doi.org/10.1016/S0044-8486(00)00364-1
Gómez-Díaz, E., González-Solís, J., 2007. Geographic assignment of seabirds to their origin: combining
morphologic, genetic, and biogeochemical analyses. Ecological Applications 17, 1484–1498.
https://doi.org/10.1890/06-1232.1
Gray, M.A., Cunjak, R.A., Munkittrick, K.R., 2004. Site fidelity of slimy sculpin (Cottus cognatus):
insights from stable carbon and nitrogen analysis. Canadian Journal of Fisheries and Aquatic
Sciences 61, 1717–1722. https://doi.org/10.1139/f04-108
Hall, L.A., Beissinger, S.R., 2017. Inferring the timing of long-distance dispersal between Rail
metapopulations using genetic and isotopic assignments. Ecological Applications 27, 208–218.
https://doi.org/10.1002/eap.1432
Hanski, I., 1998. Metapopulation dynamics. Nature 396, 41–49. https://doi.org/10.1038/23876
Hansson, S., Hobbie, J.E., Elmgren, R., Larsson, U., Fry, B., Johansson, S., 1997. The stable nitrogen
isotope ratio as a marker of food-web interactions and fish migration. Ecology 78, 2249–2257.
https://doi.org/10.1890/0012-9658(1997)078[2249:TSNIRA]2.0.CO;2
Hartl, D.L., Clark, A.G., 2007. Principles of population genetics, 4th ed. ed. Sinauer Associates,
Sunderland, Mass.
Heggenes, J., Qvenild, T., Stamford, M.D., Taylor, E.B., 2006. Genetic structure in relation to movements
in wild European grayling (Thymallus thymallus) in three Norwegian rivers. Canadian Journal of
Fisheries and Aquatic Sciences 63, 1309–1319. https://doi.org/10.1139/f06-028
Hobson, K.A., 1999. Tracing origins and migration of wildlife using stable isotopes: a review. Oecologia
120, 314–326. https://doi.org/10.1007/s004420050865
Jensen, L.F., Hansen, M.M., Carlsson, J., Loeschcke, V., Mensberg, K.-L.D., 2006. Spatial and temporal
genetic differentiation and effective population size of brown trout (Salmo trutta, L.) in small
Danish rivers. Conservation Genetics 6, 615–621. https://doi.org/10.1007/s10592-005-9014-8
Junker, J., Peter, A., Wagner, C.E., Mwaiko, S., Germann, B., Seehausen, O., Keller, I., 2012. River
fragmentation increases localized population genetic structure and enhances asymmetry of
dispersal in bullhead (Cottus gobio). Conservation Genetics 13, 545–556.
https://doi.org/10.1007/s10592-011-0306-x
Keeley, E.R., Parkinson, E.A., Taylor, E.B., 2007. The origins of ecotypic variation of rainbow trout: a test
of environmental vs. genetically based differences in morphology. Journal of Evolutionary Biology
20, 725–736. https://doi.org/10.1111/j.1420-9101.2006.01240.x
95
Kelly, J.F., Ruegg, K.C., Smith, T.B., 2005. Combining isotopic and genetic markers to identify breeding
origins of migrant birds. Ecological Applications 15, 1487–1494. https://doi.org/10.1890/04-1704
Kennedy, B.P., Chamberlain, C.P., Blum, J.D., Nislow, K.H., Folt, C.L., 2005. Comparing naturally
occurring stable isotopes of nitrogen, carbon, and strontium as markers for the rearing locations of
Atlantic salmon (Salmo salar). Canadian Journal of Fisheries and Aquatic Sciences 62, 48–57.
https://doi.org/10.1139/f04-184
Koenig, W.D., Van Vuren, D., Hooge, P.N., 1996. Detectability, philopatry, and the distribution of
dispersal distances in vertebrates. Trends in Ecology & Evolution 11, 514–517.
https://doi.org/10.1016/S0169-5347(96)20074-6
Kotsiantis, S.B., 2007. Supervised machine learning: a review of classification techniques. Informatica
249–268.
Liaw, A., Wiener, M., 2002. Classification and regression by randomForest. R news 18–22.
Manel, S., Berthier, P., Luikart, G., 2002. Detecting Wildlife poaching: identifying the origin of individuals
with bayesian assignment tests and multilocus genotypes. Conservation Biology 16, 650–659.
https://doi.org/10.1046/j.1523-1739.2002.00576.x
Manel, S., Gaggiotti, O., Waples, R., 2005. Assignment methods: matching biological questions with
appropriate techniques. Trends in Ecology & Evolution 20, 136–142.
https://doi.org/10.1016/j.tree.2004.12.004
Martin, J., Bareille, G., Berail, S., Pecheyran, C., Daverat, F., Bru, N., Tabouret, H., Donard, O., 2013.
Spatial and temporal variations in otolith chemistry and relationships with water chemistry: a useful
tool to distinguish Atlantic salmon Salmo salar parr from different natal streams: Otolith chemistry
of Salmo salar. Journal of Fish Biology 82, 1556–1581. https://doi.org/10.1111/jfb.12089
Maxwell, S.E., Kelley, K., Rausch, J.R., 2008. Sample size planning for statistical power and accuracy in
parameter estimation. Annual Review of Psychology 59, 537–563.
https://doi.org/10.1146/annurev.psych.59.103006.093735
McCarthy, I.D., Waldron, S., 2000. Identifying migratorySalmo trutta using carbon and nitrogen stable
isotope ratios. Rapid Communications in Mass Spectrometry 14, 1325–1331.
https://doi.org/10.1002/1097-0231(20000815)14:15<1325::AID-RCM980>3.0.CO;2-A
Moore, S.A., Bronte, C.R., 2001. Delineation of sympatric morphotypes of lake trout in Lake Superior.
Transactions of the American Fisheries Society 130, 1233–1240. https://doi.org/10.1577/1548-
8659(2001)130<1233:DOSMOL>2.0.CO;2
Morin, P.A., Martien, K.K., Taylor, B.L., 2009. Assessing statistical power of SNPs for population
structure and conservation studies. Molecular Ecology Resources 9, 66–73.
https://doi.org/10.1111/j.1755-0998.2008.02392.x
Morrissey, M.B., Ferguson, M.M., 2011. Individual variation in movement throughout the life cycle of a
stream-dwelling salmonid fish: Brook charr movement. Molecular Ecology 20, 235–248.
https://doi.org/10.1111/j.1365-294X.2010.04921.x
Nielsen, E.E., Hemmer-Hansen, J., Larsen, P.F., Bekkevold, D., 2009. Population genomics of marine
fishes: identifying adaptive variation in space and time. Molecular Ecology 18, 3128–3150.
https://doi.org/10.1111/j.1365-294X.2009.04272.x
Paetkau, D., Calvert, W., Stirling, I., Strobeck, C., 1995. Microsatellite analysis of population structure in
Canadian polar bears. Molecular Ecology 4, 347–354. https://doi.org/10.1111/j.1365-
294X.1995.tb00227.x
Paetkau, D., Slade, R., Burden, M., Estoup, A., 2004. Genetic assignment methods for the direct, real-time
estimation of migration rate: a simulation-based exploration of accuracy and power. Mol. Ecol. 13,
55–65.
Pakkasmaa, S., Piironen, J., 2000. Water velocity shapes juvenile salmonids. Evolutionary Ecology 14,
721–730. https://doi.org/10.1023/A:1011691810801
96
Park, P.J., Aguirre, W.E., Spikes, D.A., Miyazaki, J.M., 2013. Landmark-based geometric morphometrics:
what fish shapes can tell us about fish evolution. Tested Studies for Laboratory Teaching 361–
371.
Perrier, C., Daverat, F., Evanno, G., Pécheyran, C., Bagliniere, J.-L., Roussel, J.-M., 2011. Coupling
genetic and otolith trace element analyses to identify river-born fish with hatchery pedigrees in
stocked Atlantic salmon (Salmo salar) populations. Canadian Journal of Fisheries and Aquatic
Sciences 68, 977–987. https://doi.org/10.1139/f2011-040
Peterson, B.J., Fry, B., 1987. Stable Isotopes in Ecosystem Studies. Annual Review of Ecology and
Systematics 18, 293–320. https://doi.org/10.1146/annurev.es.18.110187.001453
Raeymaekers, J.A.M., Maes, G.E., Geldof, S., Hontis, I., Nackaerts, K., Volckaert, F.A.M., 2008. Modeling
genetic connectivity in sticklebacks as a guideline for river restoration. Evolutionary Applications
1, 475–488. https://doi.org/10.1111/j.1752-4571.2008.00019.x
Rohlf, F.J., 2015. The tps series of software. Hystrix, the Italian Journal of Mammalogy.
https://doi.org/10.4404/hystrix-26.1-11264
Saint-Pé, K., Blanchet, S., Tissot, L., Poulet, N., Plasseraud, O., Loot, G., Veyssière, C., Prunier, J.G.,
2018. Genetic admixture between captive-bred and wild individuals affects patterns of dispersal in
a brown trout (Salmo trutta) population. Conservation Genetics 19, 1269–1279.
https://doi.org/10.1007/s10592-018-1095-2
Smith, S.J., Campana, S.E., 2010. Integrated stock mixture analysis for continous and categorical data, with
application to genetic–otolith combinations. Canadian Journal of Fisheries and Aquatic Sciences
67, 1533–1548. https://doi.org/10.1139/F10-078
Smith, S.M., Maughan, P.J., 2015. SNP genotyping using KASPar assays, in: Batley, J. (Ed.), Plant
Genotyping. Springer New York, New York, NY, pp. 243–256. https://doi.org/10.1007/978-1-
4939-1966-6_18
Stelkens, R.B., Jaffuel, G., Escher, M., Wedekind, C., 2012. Genetic and phenotypic population divergence
on a microgeographic scale in brown trout: microgeographic population differentiation in brown
trout. Molecular Ecology 21, 2896–2915. https://doi.org/10.1111/j.1365-294X.2012.05581.x
Svanbäck, R., Eklöv, P., 2006. Genetic variation and phenotypic plasticity: Causes of morphological and
dietary variation in Eurasian perch. Evolutionary ecology research 37–49.
Timi, J.T., Luque, J.L., Sardella, N.H., 2005. Parasites of Cynoscion guatucupa along South American
Atlantic coasts: evidence for stock discrimination. Journal of Fish Biology 67, 1603–1618.
https://doi.org/10.1111/j.1095-8649.2005.00867.x
Waples, R.S., Gaggiotti, O., 2006. What is a population? An empirical evaluation of some genetic methods
for identifying the number of gene pools and their degree of connectivity: what is a population?
Molecular Ecology 15, 1419–1439. https://doi.org/10.1111/j.1365-294X.2006.02890.x
Ward, A.J.W., Webster, M.M., Hart, P.J.B., 2006. Intraspecific food competition in fishes. Fish and
Fisheries 7, 231–261. https://doi.org/10.1111/j.1467-2979.2006.00224.x
Yen, J., 1999. Quantitative analysis of movement: measuring and modeling population redistribution in
animals and plants. Peter Turchin. The Quarterly Review of Biology 74, 240–241.
https://doi.org/10.1086/393125
Zellmer, A.J., 2018. Microgeographic morphological variation across larval wood frog populations
associated with environment despite gene flow. Ecology and Evolution 8, 2504–2517.
https://doi.org/10.1002/ece3.3829
97
Supplementary material for Chapter II
Appendix S2.1: Morphological landmarks used to calculate relative warps
Example of two relative warps (4 and 6), which represent axes of body shape deformation.
98
Appendix S2.2: pairwise genetic differentiation between sample sites for each river basin
We computed site uniqueness as the average of pairwise FST values observed between a deme and
all other demes (Coleman, Weeks, & Hoffmann, 2013; Hedrick, 2005; Prunier et al. 2018) using
the adegenet R-package (Jombart et al. 2010), to assess whether higher genetic differentiation
would indeed enable better discrimination between populations, and therefore increase assignment
accuracy and power.
A : Aude pairwise Fst
AGUSou AIGPou AIGSou ARTLau AUDCar AUDFar AUDFou AUDNen AUDPuy AUDSer BAIPas BRUMij BRUUss CAMSau QUEMas QUERia ROQSau
AGUSou 0.000 0.053 0.022 0.075 0.052 0.024 0.032 0.040 0.038 0.025 0.522 0.063 0.039 0.036 0.021 0.020 0.140
AIGPou 0.053 0.000 0.047 0.059 0.017 0.023 0.029 0.019 0.104 0.022 0.702 0.057 0.032 0.070 0.071 0.051 0.284
AIGSou 0.022 0.047 0.000 0.072 0.047 0.023 0.034 0.037 0.038 0.029 0.524 0.067 0.042 0.044 0.022 0.024 0.143
ARTLau 0.075 0.059 0.072 0.000 0.048 0.041 0.038 0.045 0.118 0.045 0.708 0.080 0.062 0.080 0.091 0.067 0.293
AUDCar 0.052 0.017 0.047 0.048 0.000 0.017 0.020 0.013 0.106 0.021 0.699 0.041 0.027 0.063 0.070 0.050 0.285
AUDFar 0.024 0.023 0.023 0.041 0.017 0.000 0.009 0.011 0.065 0.010 0.597 0.032 0.015 0.032 0.036 0.022 0.196
AUDFou 0.032 0.029 0.034 0.038 0.020 0.009 0.000 0.013 0.074 0.010 0.618 0.028 0.020 0.029 0.046 0.027 0.212
AUDNen 0.040 0.019 0.037 0.045 0.013 0.011 0.013 0.000 0.088 0.013 0.661 0.035 0.020 0.046 0.057 0.040 0.245
AUDPuy 0.038 0.104 0.038 0.118 0.106 0.065 0.074 0.088 0.000 0.062 0.364 0.117 0.086 0.056 0.022 0.034 0.067
AUDSer 0.025 0.022 0.029 0.045 0.021 0.010 0.010 0.013 0.062 0.000 0.588 0.031 0.018 0.027 0.035 0.020 0.191
BAIPas 0.522 0.702 0.524 0.708 0.699 0.597 0.618 0.661 0.364 0.588 0.000 0.710 0.650 0.596 0.415 0.479 0.246
BRUMij 0.063 0.057 0.067 0.080 0.041 0.032 0.028 0.035 0.117 0.031 0.710 0.000 0.025 0.061 0.071 0.044 0.293
BRUUss 0.039 0.032 0.042 0.062 0.027 0.015 0.020 0.020 0.086 0.018 0.650 0.025 0.000 0.041 0.052 0.031 0.240
CAMSau 0.036 0.070 0.044 0.080 0.063 0.032 0.029 0.046 0.056 0.027 0.596 0.061 0.041 0.000 0.036 0.025 0.197
QUEMas 0.021 0.071 0.022 0.091 0.070 0.036 0.046 0.057 0.022 0.035 0.415 0.071 0.052 0.036 0.000 0.012 0.088
QUERia 0.020 0.051 0.024 0.067 0.050 0.022 0.027 0.040 0.034 0.020 0.479 0.044 0.031 0.025 0.012 0.000 0.120
ROQSau 0.140 0.284 0.143 0.293 0.285 0.196 0.212 0.245 0.067 0.191 0.246 0.293 0.240 0.197 0.088 0.120 0.000
B : Seuge pairwise Fst
BUIPie CLATou LAVRiv LAVSuc PONAmo PONBom PONCha PONTet PONTis SERSau SEUCch SEUCou SEUCro SEUFag SEURod SEUSau
BUIPie 0.000 0.191 0.166 0.099 0.066 0.062 0.094 0.151 0.083 0.193 0.022 0.110 0.134 0.024 0.100 0.105 CLATou 0.191 0.000 0.064 0.047 0.052 0.085 0.037 0.171 0.044 0.069 0.096 0.044 0.066 0.071 0.038 0.042 LAVRiv 0.166 0.064 0.000 0.018 0.061 0.104 0.016 0.165 0.017 0.047 0.142 0.020 0.049 0.110 0.017 0.021 LAVSuc 0.099 0.047 0.018 0.000 0.060 0.110 0.012 0.131 0.012 0.027 0.137 0.017 0.045 0.112 0.008 0.014 PONAmo 0.066 0.052 0.061 0.060 0.000 0.045 0.051 0.071 0.045 0.032 0.082 0.072 0.091 0.059 0.054 0.060 PONBom 0.062 0.085 0.104 0.110 0.045 0.000 0.102 0.051 0.088 0.068 0.065 0.125 0.157 0.059 0.107 0.116 PONCha 0.094 0.037 0.016 0.012 0.051 0.102 0.000 0.116 0.009 0.021 0.129 0.019 0.044 0.103 0.009 0.012 PONTet 0.151 0.171 0.165 0.131 0.071 0.051 0.116 0.000 0.106 0.172 0.066 0.141 0.127 0.061 0.126 0.132 PONTis 0.083 0.044 0.017 0.012 0.045 0.088 0.009 0.106 0.000 0.021 0.114 0.021 0.047 0.090 0.012 0.017 SERSau 0.193 0.069 0.047 0.027 0.032 0.068 0.021 0.172 0.021 0.000 0.075 0.033 0.046 0.055 0.022 0.029 SEUCch 0.022 0.096 0.142 0.137 0.082 0.065 0.129 0.066 0.114 0.075 0.000 0.152 0.187 0.015 0.136 0.145 SEUCou 0.110 0.044 0.020 0.017 0.072 0.125 0.019 0.141 0.021 0.033 0.152 0.000 0.049 0.117 0.017 0.014 SEUCro 0.134 0.066 0.049 0.045 0.091 0.157 0.044 0.127 0.047 0.046 0.187 0.049 0.000 0.149 0.040 0.042 SEUFag 0.024 0.071 0.110 0.112 0.059 0.059 0.103 0.061 0.090 0.055 0.015 0.117 0.149 0.000 0.107 0.116 SEURod 0.100 0.038 0.017 0.008 0.054 0.107 0.009 0.126 0.012 0.022 0.136 0.017 0.040 0.107 0.000 0.012 SEUSau 0.105 0.042 0.021 0.014 0.060 0.116 0.012 0.132 0.017 0.029 0.145 0.014 0.042 0.116 0.012 0.000
99
C: Boxplots of uniqueness (mean pairwise Fst values) for the Aude and the Seuge Rivers. Black
lines represent the median (higher in the Seuge whereas the mean is lower).
100
Appendix S2.3: Isotoscapes for periphyton, baetis and trout
A: Aude:
101
Seuge:
102
Chapter III. Captive-bred ancestry affects spatial patterns of
genetic diversity and differentiation in Brown trout (Salmo
trutta) populations
Saint-Pé Keoni1*
, Laurence Tissot3, Nicolas Poulet
4, Jérôme G. Prunier
1 and Simon Blanchet
1,2*
1 Centre National de la Recherche Scientifique (CNRS), Université Paul Sabatier (UPS); Station
d‘Ecologie Théorique et Expérimentale, UMR 5321, F-09200 Moulis, France
2 CNRS, UPS, École Nationale de Formation Agronomique (ENFA) ; UMR 5174 EDB
(Laboratoire Évolution & Diversité Biologique), 118 route de Narbonne, F-31062 Toulouse cedex
4, France
3 EDF R&D LNHE - Laboratoire National d‘Hydraulique et Environnement, 6 quai Watier, 78401
Chatou Cedex, France
4 Pôle écohydraulique AFB-IMT, allée du Pr Camille Soula, 31400 Toulouse France
Key words: Genetic admixture, SNPs, diversity and differentiation patterns, brown trout,
individual assignment, captive breeding, stocking, hatchery.
103
Summary
Intraspecific patterns of diversity and differentiation are key features for populations.
Understanding how human induced perturbations affect these patterns is therefore a crucial
question. Among the numerous pressures which directly or indirectly affect patterns of diversity
and differentiation, stocking natural populations with captive-bred strains is a common practice
worldwide.
Using 1738 individuals genotyped at 192 SNP loci, we quantified and described the spatial
distribution of captive-bred ancestry in brown trout (Salmo trutta) populations from four
watershed having been stocked using Bayesian clustering. We then tested the effect of captive-
bred ancestry on the spatial distribution of genetic diversity and differentiation. Individuals with a
strong assignment to the captive strain occurred almost exclusively in upper reaches of the basins.
We found that captive-bred genotypes increased diversity and differentiation, and that patterns
such as downstream increase in genetic diversity and higher differentiation upstream were
reversed in the case of higher captive-bred genotypes occurring at the level of populations.
The heterogeneous distribution of captive-bred genotypes, even stocked many generations ago, as
well as their effect on diversity and differentiation patterns, is a crucial facet of eco evolutionary
dynamics which is far overlooked in most studies on spatial processes in stocked species. Indeed,
as natural patterns of diversity and differentiation reflect important processes driving the dynamics
of populations, understanding how human induced changes can affect them has great implications.
104
Introduction
Describing and understanding spatial patterns of biodiversity is a central question in ecology,
evolution and conservation sciences (Chave 2013). Large spatial scale patterns of interspecific
diversity have been widely studied, such as distribution of species‘ richness across latitudinal
ranges (Gaston 2000). It has further been shown that intraspecific diversity follows the same
global patterns as interspecific diversity, with higher diversity in the tropics and a decrease toward
the poles (Miraldo et al. 2016). At more local scales (i.e. within landscapes) many studies have
assessed patterns of diversity, such as functional diversity (Raffard et al. 2017) and genetic
diversity (Charlesworth et al. 1997; Manel et al. 2003; Paz-Vinas et al. 2015). Describing and
understanding small scale patterns of intraspecific diversity is of utmost importance for designing
proper conservation strategies (Paz-Vinas et al. 2018) given that intraspecific diversity is the first
biodiversity facet being affected by human pressures (Mimura et al. 2017), and that it constitutes
the evolutionary potential of species (Franklin and Frankham 1998).
In river ecosystems, intraspecific patterns of genetic diversity and differentiation have been widely
studied (Labonne et al. 2008; Paz-Vinas et al. 2015). River ecosystems are characterized by their
tree-like geometric branching pattern (dendritic network; Benda et al. 2004; Campbell Grant et al.
2007), and strongly structured by elevation, making water flow unidirectional. These
characteristics were shown to modulate patterns of genetic diversity and differentiation. Indeed,
these landscapes strongly constrain movements of individuals and hence dispersal, with for
instance smaller and less connected patches upstream (therefore experiencing more drift;
Raeymaekers et al. 2008; Carrara et al. 2014) and asymmetrical gene flow due to the downstream
direction of the water flow (Morrissey and de Kerckhove 2009). Moreover, colonization history
(which usually occurred from downstream towards upstream reaches; Cyr and Angers 2011) was
also shown to be crucial in shaping these patterns in river ecosystems.
Altogether, these mechanisms have been shown to lead to recurrent spatial patterns consisting in a
downstream increase in genetic diversity (hereafter DIGD, Kikuchi et al. 2011; Alp et al. 2012;
Torterotot et al. 2014; Paz-Vinas et al. 2015) and an upstream increase in differentiation. Although
numerous and thorough, studies on these patterns have to our knowledge rarely taken into account
the fact that anthropologic pressures can alter patterns of intraspecific diversity and differentiation
105
(but see Prunier et al. 2018) by modifying their underlying mechanisms (Mimura et al. 2017).
River fragmentation by weirs or dams for instance, generally reduces gene flow (Keller and
Largiadèr 2003; Raeymaekers et al. 2008; Blanchet et al. 2010) which may reinforce DIGD and
upstream genetic differentiation (Blanchet et al. 2010; Faulks et al. 2011). Conversely, a human-
induced reduction in effective population size in lower reaches (for instance due to pollution or
habitat destruction; Almodóvar et al. 2012; Comte et al. 2013), should reduce genetic diversity of
these downstream demes (Ellstrand and Elam 1993), and hence cancel or even invert DIGD and
increase downstream genetic differentiation.
Moreover, some pressures are more insidious, and may not affect spatial patterns through
population size and connectivity directly. Among these pressures, stocking of captive-bred
individuals is a widespread practice used to sustain and/or enhance populations to improve
recreational fishing (Borsuk et al. 2006; Hansen et al. 2009; Burnside et al. 2016). Because they
are from different origin (lineage, mainly from north Europe AT), and because of unnatural
selection pressures and small population sizes, captive-bred individuals strongly differ genetically
from wild ones (Frankham et al. 1986; Blanchet et al. 2008; Christie et al. 2016). The genetic
effects of stocking have thus been a concern for both scientists and river managers, in particular
admixture between wild and captive-bred strains (Cagigas et al. 1999; Heggenes et al. 2002;
Perrier et al. 2013a). The differences between strains and the potential genetic effects of stocking
on populations suggest that this practice could modify natural patterns of intraspecific diversity.
First, direct impacts of stocking on genetic diversity and differentiation (Moran et al. 2005)
should, depending on the location of stocking events, modify the natural patterns. Moreover, as for
instance captive-bred were shown to disperse differently than wild individuals, and occupy higher
reaches (Saint-Pé et al. 2018), we can expect that their location in a river system will strongly
influence natural patterns, by homogenizing structure and increasing diversity upstream.
The Brown trout (Salmo trutta), as most salmonids, naturally exhibits strong patterns of genetic
diversity and differentiation at small spatial scales (Aurelle and Berrebi 2001, Swatdipong et al.
2010; Vøllestad et al. 2012), which have been studied widely because of high interest for river
managers and researchers (Berrebi 2015). As it is highly associated to human interests (Mills
1989; Butler et al. 2009), the Brown trout has been domesticated since the 19th
century (Laikre et
al. 1999), and hatchery strains have been largely used to sustain wild populations worldwide
106
(Elliott 1994; Berrebi et al. 2000; Bohling 2016; Lobón-Cervía and Sanz 2017). However, the
effects of this practice were rapidly shown to be highly concerning, because of its negative effects
on natural populations (Hansen 2002; Araki and Schmid 2010; Christie et al. 2012). Therefore,
assessing the effect of passed stocking activities at a multi-river shed level, with the aim of
describing a general effect of this practice on expected spatial patterns of genetic diversity and
differentiation is of crucial importance for both scientists and managers.
Here, we tested the effect of stocking and admixture between wild and captive-bred Brown trout
(Salmo trutta) on spatial patterns of genetic diversity and differentiation in replicated river
systems. By combining extensive river scale sampling design and a series of tools derived from
population genetics approaches, we specifically aimed at (i) quantifying captive-bred ancestry of
supposedly native populations (ii) testing how captive-bred ancestry was spatially distributed
within a river catchment and (iii) testing how captive-bred ancestry affects the relations between
genetic diversity/differentiation and distance to the river mouth in four independent river basins.
We expect that, captive-bred individuals and their descendants are not distributed homogeneously
throughout river basins, and are mostly present in upstream reaches, where the density of native
counterparts is generally lower resulting in lower intraspecific competition and prior effects
(Weber and Fausch 2003; Saint-Pé et al. 2018). Partly due to this effect, we therefore expect that
spatial patterns of differentiation and diversity will be different according to the degree of
―nativeness‖ of individuals/populations. We expect that DIGD for instance will be inexistent (or
even inverted) when considering captive-bred individuals and their descendants. Finally, as these
strains tend to live upstream, in unconnected reaches although they are from a ―homogeneous
cluster‖, we thus expect that differentiation in these upper reaches in which it is naturally expected
to be very high, will be lower because of the common origin and ―sink effect‖ of hatchery
genotypes.
Methods
Study area
We focused on four independent French river basins: the Aude, the Ône, the Seuge and the Roya
Rivers (Figure 1), in which the fish community is dominated by brown trout. These basins vary
considerably in terms of geography and environmental conditions, and in stocking activity (Table
107
1). These four river basins, located in southern France, flow from three different mountain ranges:
the Ône and the Aude Rivers flow from the Pyrenees Mountains, the Seuge River from the Massif
Central, and the Roya River flows from the Alps. They naturally harbor two different lineages of
brown trout, due to their oceanic outlet (Bernatchez 2001): Atlantic (AT) lineage in the Ône and
the Seuge Rivers, Mediterranean (ME) lineage for the Aude and the Roya Rivers. The study areas
concern the upper reaches of the catchments: above the lowest sampling sites, the basins drain in
average 205 km2 (240 km
2 for the Aude catchment, 155 km
2 for the Ône, 360 km
2 for the Roya,
and 90 km2 for the Seuge basin).
Figure 3.1: Map of the four river basins studied. The Seuge and Ône Rivers are part of the Atlantic lineage,
whereas the Aude and the Roya Rivers are part of the Mediterranean lineage. Sampling sites = black dots
108
Ône Seuge Aude Roya
Outlet Atlantic Atlantic Mediterranean Mediterranean
Mountain range Pyrenees Massif Central Pyrenees Alps
Elevation range (m) 630-1400 960-1250 380-1420 300-1550
Surface drained by the basin above lowest sampling point (km2)
240 180 155 360
Flow source Snow-rain Snow-rain Snow-rain Snow-glacier-rain
Hatcheries used in the basins
Soueich Vourzac, Federation of Lozere
Gesse, Fagolle Roquebillère, Italian Atlantic hatcheries
Hatcheries successfully sampled
Soueich Vourzac Fagolle Roquebillère + Soueich
Stocking range 1972-1999 1970’s-1995 1970’s-2000 1970’s-present
Stocking intensity ++ - + +++
Table 3.1: Features for each basin, presenting the outlet (Atlantic or Mediterranean), the mountain range,
elevation range, basin surface, flow source, the hatcheries used (of which we know), hatcheries sampled,
time range during which stocking occurred, and intensity of stocking.
General stocking practices in France
For more than a century (and especially in the last 50 years), stocking practices have led to
massive introductions of hatchery-reared AT (majority of Danish origin) trout (Laikre et al. 1999),
even in rivers naturally inhabited by ME trout (Krieg and Guyomard 1985). However, because this
had tremendous consequences on native populations (Largiader and Scholl 1996; Berrebi et al.
2000; Caudron et al. 2006), French managers changed their stocking practices since the early
2000s: stocking was either stopped or shifted toward the use of more local strains (e.g. through the
use of ME strains for Mediterranean rivers; Caudron et al. 2006) in many stocking reaches.
Nonetheless, in France, data on stocking is generally scarce, and only qualitative data about the
source of stoking is properly informed by local managers. Regarding the four target rivers, we
109
provide in Table 1 the information that we were able to collect from formal discussions with
angling departments and local managers about stocking history, intensity and strategy (see also
hereafter).
Sampling hatcheries according to stocking practices
Quantifying the impacts of stocking and the potential for admixture between wild and stocked fish
requires to genetically characterizing the stocked fish. Here, we focused on the hatchery strains
having been used in the last 10-30 years in the basins since information anterior to this period is
inexistent. For three rivers (the Ône, Seuge and Aude Rivers), stocking practices were officially
stopped in the early 2000‘s (unpublished data, Table 1). In the Ône River, stocking was done from
an Atlantic strain local trout hatchery (Soueich, origin of fish is a mix between local rivers and
Danish strains) that is administrated by the departmental angling association (Fédération
Départementale pour la Pêche et la Protection des Milieux Aquatiques de Haute Garonne). The
Aude River was stocked with two different Atlantic strains from local hatcheries: Gesse and
Fagolle. We did not manage to get samples of the Gesse hatchery as it has ceased its activity some
years ago. These hatcheries were used until the 1980ies, and stocking has officially been stopped
since then. The Seuge River was stocked with fish from the hatchery of the departmental angling
association of Lozère (Fédération Départementale pour la Pêche et la Protection des Milieux
Aquatiques de Lozère) until 1989, with Atlantic strain individuals (the exact origin being
unknown). After contacting many actors involved, we were unable to obtain samples from this
hatchery that does no longer exist. Nonetheless, we collected samples from a small local hatchery
derived from it, which was also used to stock this basin until 1989 (Pisciculture des eaux du
Vourzac). Finally, in the Roya River, the story is more complex as it is managed according to two
practices. In the downstream sections of the river (below Fontan, ―ROY-Fon‖, see Figure 1)
stocking practices have been stopped recently (2007), whereas in the upstream sections stocking
practices are still ongoing. This downstream part of the river has been stocked by fish from
multiple hatcheries: first by the Fontan hatchery (Atlantic strain, unknown origin, stopped in
2006), then by the Roquebilière hatchery (Mediterranean strain, origin unknown) administrated by
the departmental angling association of Alpes-Maritimes (Fédération Départementale pour la
Pêche et la Protection des Milieux Aquatiques des Alpes-Maritimes). Stocking was officially done
until 2007 with this hatchery. Contrastingly, the upper reaches are still stocked with Atlantic
110
strains (unknown origin, most likely from Italian hatcheries), but their tracking is extremely
complex, as it exists little exchange between local managers and legal authorities. Roquebilière
hatchery has been successfully sampled, but Fontan samples were not available. We therefore used
Soueich as an Atlantic hatchery to represent stocking in this basin (note that using Soueich or one
of the other Atlantic hatcheries did not change the results qualitatively and quantitatively). As a
result of information gathered (Table 1), we considered the Roya River as having been (and being)
the most potentially impacted by stocking practices, particularly in its upper reach. According to
interviews with local managers, we considered that the second most potentially impacted (number
and duration of stocking events) is the One River, then the Aude River and finally the Seuge River
that have been only weakly impacted (Table 1).
For each basin, we therefore sampled 30 fish from each of the trout hatcheries used, which were
available, in order to genetically characterize the captive-bred strain, and hence quantify genetic
admixture with captive-bred trout. In Table 1, we give the hatcheries used (which we know of),
and the ones we were able to obtain samples for.
Field sampling
Brown trout were sampled using a single-pass electric-fishing approach from 71 sites in total, with
an aim of catching 30 individuals per site. The number of sites per river basin varied from 16
(Seuge River) to 21 (Roya River) so as to cover the whole river network (all main tributaries and
the upstream-downstream gradient of the main stem) of each river (Figure 1, Appendix S1). We
focused on small-bodied individuals during sampling in order to focus mainly on individuals from
one age class (1 year-old). In total, we captured 1738 individuals, with sample sizes ranging from
6 to 30 (Appendix S1 for details). Each individual was measured (total length in mm, see
Appendix S1 for more details on the size distribution), and a pelvic fin clip was taken for the
genetic analysis. All individuals were released alive to their original sampling site.
Genotyping
Individual multilocus genotypes were obtained for 162 polymorphic SNPs markers (Saint-Pé et
al., 2019), using the KASP technology performed by LGC Genomics® (Smith and Maughan
2015). These markers were developed from a high-density linkage map (Leitwein et al. 2017), and
111
filtered to be highly polymorphic, evenly spread and spaced on the linkage map, and present in
both AT and ME brown trout lineages (see Saint-Pé et al., 2019 for details). Five ancestry
informative SNPs used in previous studies to distinguish individuals (and infer admixture) from
the AT and ME lineages (OMM1164, OMM1105, OMM1154, Str541INRA, Str591INRA; Estoup
et al. 2000; Caudron et al. 2006, 2012), were included in the array.
Quantification of captive-bred ancestry and genetic diversity and differentiation
For each river independently, we performed genetic clustering using STRUCTURE 2.3.1
(Pritchard et al. 2000) with the admixture model and the correlated allele frequency model,
without prior population information. Runs were performed with a burn-in period of 200.000 and
200.000 subsequent MCMC repetitions. The number K of clusters ranged from 1 to 10 and 5 runs
were performed for each value. We checked MCMC convergence making sure the alpha plots
showed no substantial fluctuation before the end of the burn-in. Log-likelihood plots and ΔK
statistics were obtained via the Evanno‘s method (Evanno et al. 2005) using STRUCTURE
HARVESTER (Earl and vonHoldt 2012), and were used to infer the optimal K-value. Twenty runs
were then performed with this optimal K-value and the ten best runs (the ones with highest
LnP(D) values) were compiled using the Greedy algorithm from CLUMPP (Jakobsson and
Rosenberg 2007) to get final averaged individual Q-values. Individuals were assigned to the
cluster for which their Q-value was highest, and provided this value was higher than 0.7.
Graphical displays of STRUCTURE plots were generated using DISTRUCT software (Rosenberg
2003) with the membership of each individual representing the mean membership over the
replicate runs. We directly used the probability of assignment (individual Q-value) to the cluster
containing all individuals from the trout hatcheries used to stock each basin (see the Results
section) as an estimate of individual captive-bred ancestry (Hansen et al. 2001; Vähä and Primmer
2005; Valiquette et al. 2014). This probability thus varies from 0 to 1 and the higher the score, the
more the individual carries captive-bred ancestry (individuals with a score close to 1 are likely
captive-bred fish released or pure descendants of captive-bred fish). In the case of the Roya basin,
as there were two distinct hatchery strains used to stock (ME and AT), we used two steps to
identify ancestry origin of captive-bred ancestry: first, we assessed the percentage of AT ancestry,
then we assessed, for individuals considered as ―Mediterranean‖, the probability of assignment to
the ME hatchery of Roquebillière.
112
Genetic diversity within each of the sampling site was estimated over all loci by computing
unbiased expected heterozygosity (He), while genetic differentiation was assessed by computing
the global Fst for each site using the adegenet R package (Jombart 2008). This later measure of
differentiation indicates to which extent a site is genetically unique/differentiated compared to all
other sites form the river.
Statistical analyses
Spatial patterns of captive-bred ancestry. In order to describe the spatial structuration of strains in
river basins, we tested the relationship between site riverine distance to the river mouth (calculated
with STARS ArcGis package, Peterson and Hoef 2014) and proportion of captive-bred ancestry
(averaged at the site level over all sampled individuals), using a beta regression model (from the
betareg R package, as the proportion of captive-bred ancestry is comprised between 0 and 1) that
also included a two-way interaction term between the identity of the river basin and the distance to
the mouth to test if the relationship varied between basins. Using the dredge R function, we
selected the best models based on AIC, and determined the proportion of models displaying a
ΔAIC lower than 4 in which each variable appeared.
Spatial patterns of genetic diversity and differentiation, and the effect of introgression. We tested
the effect of the proportion of captive-bred ancestry on genetic diversity across basins, running a
beta regression model in which the mean expected heterozygosity (as a measure of genetic
diversity, comprised between 0 and 1) at the site level was the response variable, while the
proportion of captive-bred ancestry, the distance to the mouth, the river basin and all resulting
two-term interactions were the predictors. In the same way, we tested the effect of captive-bred
ancestry on the spatial distribution of genetic differentiation using a linear model (lm from the glm
R package). In that case, the mean Fst (as a measure of population differentiation) at the site level
was the response variable, while the proportion of captive-bred ancestry, the distance to the mouth,
the river basin and all resulting two-term interactions were the predictors. The best models were
also selected based on AIC using dredge, and we determined the proportion of models displaying a
ΔAIC lower than 4 in which each variable appeared. A significant interaction term between the
proportion of captive-bred ancestry and site river distance to the river mouth would indicate that
113
spatial patterns of intraspecific diversity (either the DIGD or the increase in differentiation
upstream) are affected by introgression by stocked individuals.
Results
Spatial patterns of captive-bred ancestry
Captive-bred ancestry presence strongly varied across rivers (Figure 2A). At the individual level, it
ranged from 0 to 0.99, with an average of 0.28 ± 0.31 (mean ± SD) (see Figure 2A and Appendix
S2 for details). Moreover, we found that the spatial distribution of captive-bred ancestry was
heterogeneous, mostly found in the upper parts of the basins. This was statistically confirmed as
we found that a single model was best supported by the data for captive-bred ancestry (Appendix
S3A): this model comprised ―river basin‖ and ―riparian distance to the river mouth‖ as simple
terms (Table 2a), indicating that captive-bred ancestry was different in average among basins (as
shown in figure 2A), and positively related to riparian distance to the river mouth (Figure 2B).
A B
Figure 3.2: (A) Boxplots of captive-bred ancestry for each basin, averaged at the population level. (B) Plot
of captive-bred ancestry against distance to the river mouth (scaled). Line and 95% confidence interval
correspond to the fitted values of the beta regression.
114
Basin Dist-mouth Dom_anc Basin:Dist_mouth Basin: Dom_anc Dist_mouth: Dom_anc
(A) Dom_anc
(Betareg) 1 1 / 0 / /
(B) He (Betareg) 1 0.75 1 0.25 1 0.25
(C) Fst (lm) 1 0.5 1 0.83 0.33 0.67
Table 3.2: Proportion of models (with DeltaAIC<4 after using dredge) containing each explanatory
variable tested (River basin (“Basin”), riparian distance to the river mouth (“Dist-mouth”), proportion of
captive-bred ancestry (“Dom_anc”), and two way interactions (“:”) for explaining the proportion of
captive-bred ancestry (“Dom_anc”), diversity (“He”) and differentiation (“Fst”). The type of model used
is detailed in the first column: beta regression (“betareg”) or linear model (“lm”).
More precisely, in the Ône River, fish with a high proportion of captive-bred ancestry (q > 0.7)
were present in the most upstream site of the Neste d‘Oueil (―NOU-Boo‖, which has not been
stocked since 1998, although illegal non-controled stocking cannot be ruled out). The Neste de
Garin tributary (―NGA sites‖) was almost exclusively populated with fish with a high proportion
of captive-bred ancestry. Other sites were mostly populated with native fish, which were separated
in two main genetic clusters (Figure 3A). In the Seuge River, all individuals were assigned by less
than 0.3 to the captive-bred cluster, which indicated that this basin was populated with only native
fish, separated in two clusters (Wild 1, in the upper part, and Wild 2 in the downstream part, and
some sites being populated with hybrids; Figure 3B). In Mediterranean Rivers (Aude and Roya
Rivers), captive-bred ancestry was also heterogeneous spatially. The Aude basin displayed
relatively low captive-bred ancestry (Figure 3C), except in two tributaries: ―BAI-Pas‖, in which all
individuals were highly assigned to the captive-bred cluster, and ―ROQ-Sau‖ populated with
mostly hybrids between native and captive-bred strains. On the other hand, the Roya basin
displayed a very high proportion of fish with a high proportion of captive-bred ancestry. Upstream
from Fontan (―ROY-Fon‖, Figure 3D), there were virtually no fish assigned as pure native fish.
Atlantic fish are mostly present in the upper reaches, and a relatively high proportion of these fish
are hybrids between Atlantic and Mediterranean strains. These hybrids are most likely captive-
bred Mediterranean in these reaches, as no native ancestry was found. Captive-bred ancestry from
the Mediterranean hatchery was also ubiquitous in these upper reaches, and hybrids between this
strain and native individuals were also present in high proportions in the downstream reaches
(Figure 3D).
115
Figure 3.3: Maps of each river basin, with proportion of individuals assigned to each cluster (native,
domestic and hybrids clusters) at each site
Effects of captive-bred ancestry on spatial patterns of genetic diversity and differentiation
Expected heterozygosity over all loci (He) and all populations was comprised between 0.041 and
0.40 (mean = 0.18 ± 0.10), and Fis within each site ranged from -0.27 to 0.26 (0.01 ± 0.07).
Genetic differentiation (Fst) estimated across loci at the site level (―uniqueness‖) was comprised
between 0.04 and 0.53 (mean = 0.09 ± 0.06). Trout from the Roya River were –in average- the
116
most diverse genetically, whereas the highest levels of differentiation were observed in the trout
from the Aude River (Table 3 and Appendix S1 for details).
He Fis Fst
min max mean sd min max mean sd min max mean sd
Aude River 0.066 0.388 0.162 0.088 -0.129 0.262 0.032 0.086 0.063 0.534 0.114 0.112
Ône River 0.091 0.182 0.139 0.022 -0.053 0.070 0.003 0.034 0.042 0.135 0.063 0.024
Roya River 0.220 0.398 0.305 0.064 -0.029 0.072 0.023 0.031 0.048 0.229 0.091 0.043
Seuge River 0.041 0.110 0.090 0.019 -0.269 0.151 -0.035 0.112 0.045 0.112 0.069 0.020
Table 3.3: Per basin summary statistics of expected heterozygosity (He), Fis and genetic differentiation
(Fst) estimated across loci for each site
Regarding He, four models displayed ΔAIC<4 (Appendix S3B), and basin identity, captive-bred
ancestry and the interaction term between these two variables were selected in all these four
models. This indicated that He was positively linked to the proportion of captive-bred ancestry for
all rivers, although the slope of this relation significantly varied across basins (Figure 4). In
particular, the positive relationship between He and captive-bred ancestry was stronger for the
Aude and Roya Rivers, than for the Ône and Seuge Rivers where the relationship was extremely
weak (Figure 4). Moreover, three out of the four models with a ΔAIC<4 models retained the term
―riparian distance to the river mouth‖, and one model retained the interaction term between
riparian distance to the river mouth and basin identity (Appendix S3B, Table 2b). This indicated
that the relationship between He and distance from the mouth varied between river basins, overall
this relationship was negative for the Seuge River (as expected according to DIGD), but positive
for other basins (the highest slope was for the Roya River; Appendix S4). Finally, the interaction
term between captive-bred ancestry (at the site level) and distance to the river mouth was retained
in one out of the four model with ΔAIC<4 (Table 2b; Appendix S3B) in explaining He, indicating
that the relationship between He and distance from the mouth tends to vary according to the
probability of assignment to the captive-bred cluster. In particular, a visual inspection of these
links reveals that there was –as expected according to the DIGD- a general downstream increase in
expected heterozygosity for populations composed of individuals strongly assigned to the native
cluster, whereas populations composed of individuals strongly assigned to the captive-bred cluster
showed a striking increase in expected heterozygosity in upstream reaches (Figure 5a). As captive-
bred ancestry proportion was different between basins, for instance inexistent in the Seuge and
117
very high in the Roya, these latter results also thus explain why the relation between He and
riparian distances vary between basins: they interact with the level of captive-bred ancestry.
Figure 3.4: Plots of genetic diversity (expected heterozygosity, He) against proportion of captive-bred
ancestry depending on the river basin as an interaction term
Figure 3.5: Plots of genetic diversity (He) (A) and differentiation (Fst) (B) against riparian distance to the
river mouth (in m) depending on proportion of captive-bred ancestry as an interaction term
118
Regarding genetic differentiation, 6 models displayed ΔAIC<4 (Appendix S3B), and basin identity
and captive-bred ancestry were selected in all these 6 models indicating that these two variables
were extremely important to explain patterns of Fst (Table 2c). To a lesser extent distance from the
mouth was selected in half of the models displaying a ΔAIC<4 (Table 2c). Nonetheless, three
interaction terms involving these three terms were also selected in some of the most likely models
(Table 2c), indicating that these variables likely have interactive effects on Fst. Overall, genetic
differentiation tended to increase upstream (as expected), but this increase (i) tended to vary
among rivers, with the lowest slope for the Aude River, the highest for the Seuge (interaction
―basin:dist_mouth‖, Appendix S5), (ii) tended to be lower in slope for sites with a higher
proportion of captive-bred ancestry than for sites mainly containing native assigned fish
(interaction ―dist_mouth:dom_anc‖, Figure 5b). Moreover, higher Fst values were observed for
sites with a higher proportion of captive-bred ancestry.
Overall, we found that captive-bred ancestry is a strong driver of both genetic diversity and
differentiation. Indeed, in our models, this variable appears very frequently (Appendices S3A and
S3B), and is more present than distance to the river mouth, a variable which would be most
expected to drive these patterns.
Discussion
Captive-bred ancestry extent
Although stocking was stopped years ago for most basins (i.e. 6-8 trout generations ago), captive-
bred ancestry is still present, from very low percentages, 4% in the Seuge River (26 years without
stocking), to 59% in the Roya River (still heavily stocked in its upstream part nowadays). This
high variability is congruent with other studies which show that admixture rates between captive-
bred and wild trout populations are highly variable, ranging from undetectable contributions to
total replacement of native gene pools (Poteaux 1999; Cagigas et al. 1999; Perrier et al. 2013b;
Kazyak et al. 2018). Captive-bred ancestry and introgression outcomes depending on stocking
strategies and durations have been widely studied (Barbat-Leterrier et al. 1989; Largiadèr et al.
1996), and shown to be highly variable depending on the context. For instance, Almodovar et al.
119
(2006) found alarmingly high rates in some Mediterranean rivers, whereas North Atlantic
populations showed little or no introgression, which is congruent with our results.
Although not quantified statistically, we observed a tendency for captive-bred ancestry presence
and introgression to be positively associated to stocking intensity and to how recent the last
stocking events are (Martinez et al. 1993; Marie et al. 2010; Gossieaux et al. 2019) as the Seuge
River (where stocking was stopped long ago) is the least impacted whereas the Roya (still partially
heavily stocked) is the most impacted. Populations could thus recover from stocking (Post 2013;
Valiquette et al. 2014; Létourneau et al. 2018), thanks to several mechanisms: lower fitness and
survival of captive-bred individuals (Aarestrup et al. 2005; Pedersen et al. 2008), genetic drift
purging exogenous alleles (Frankham et al. 2010), competitive exclusion (Saint-Pé et al., 2018), or
difference in spawning time (Hansen et al. 2006). Fishing pressure after stocking can also be
determinant for the survival of stocked fish, as it was shown that stocked fish were more likely to
be caught (Mezzera and Largiadèr, 2005; García-Marín et al. 1998). However, although the
cessation of stocking has in some cases enabled to return to the ―natural state‖ (Hansen et al. 1995;
Almodovar et al. 2001), we believe that in an extreme case such as the Roya, post-stocking
recovery rates would be low even at long-term if stocking was ceased, as it was observed in
populations of France and Spain (Poteaux 1999; Araguas et al. 2004).
A major limitation in studies on the genetic impact of stocking, at least in southern Europe, is the
lack of reliable information in terms of activity, density, and time range (de Sostoa and Lobon-
Cervia 1989). Moreover, from our experience, we found that historical archives are scarce, and
because of numerous illegal and hidden stocking events (unconsented by Angling departments),
conclusions on the persistence of captive-bred strains are often impossible to make. Other
salmonids (e.g. brook trout; Marie et al. 2010; Gossieaux et al. 2019) have been much better
surveyed in terms of stocking, and therefore, provided the impacts of stocking are consistent
across salmonids, they potentially represent more reliable models.
Captive-bred ancestry distribution
Interestingly, for all basins, the spatial distribution of captive-bred ancestry in the watershed was
not homogeneous (consistent with Saint-Pé et al., 2018), whereas stocking activities are generally
120
done homogeneously over river basins. In all basins, captive-breeding ancestry was majorly
present in upstream reaches, with some upper sites exclusively populated with fish strongly
assigned to the captive-bred cluster, whereas in lower reaches, fish strongly assigned to the
captive-bred cluster were rare. For instance in the Ône basin, captive-bred ancestry was present in
the upper reaches of the Neste d‘Oueil (―NOU-Bou‖ site), and the Neste de Garin (―NGA‖ sites,
Figure 3). In the Roya basin, this pattern was also observed (Figure 3), as all the upper reaches
were exclusively populated with captive-bred strains, although in this basin, stocking practices are
different between upstream and downstream reaches, and are thus propbably the main cause for
this pattern. These visual patterns were confirmed statistically as we found a significant
relationship between distance to the river mouth of each site and the proportion of captive-bred
ancestry. Conditions in these upper reaches may differ significantly from those downstream in
terms of habitat availability and stability (Vannote et al. 1980; Grant et al. 2012), therefore, fish
assigned to captive-bred clusters may be favored by harsher conditions. Indeed, higher
introgression has been linked to low pH, high water temperature (Marie et al. 2012; Harbicht et al.
2014; Létourneau et al. 2018), environmental instability and smaller habitat size (Splendiani et al.
2013; White et al. 2018). There is however a counter-example of this pattern (with higher
introgression in fertile waters and stable discharge; Almodovar et al. 2006), but in all cases, there
is a spatial segregation between strains. Various mechanisms could underlie this pattern. First,
movement patterns between wild and captive-bred strains may differ (Vasemägi et al. 2005;
Finnegan and Stevens 2008; Saint-Pé et al. 2018). Second, competition with the native strain may
probably be low or inexistent in these harsher stretches, thus facilitating the settlement of non-
native strains and improving their reproductive success (Saint-Pé et al., 2018).
Effect of stocking on spatial patterns of diversity and differentiation
The distribution of allochtonous alleles introduced with stocking has modified spatial patterns of
genetic diversity and differentiation. This is consistent with Prunier et al. (2018), who found that
among various anthropogenic stressors, stocking had a strong and consistent influence on patterns
of genetic diversity and genetic differentiation. Love Stowell et al. (2015) also found that the
actual distribution of lineages was predicted by stocking pressure and source, but not by which
lineage was historically native in a Cutthroat population (Oncorhynchus clarkii). Stocking
121
therefore has a tremendous effect on wild populations, which foreshadows concerning
implications.
Captive-bred fish generally harbor lower genetic diversity than wild ones (Hansen et al. 2006;
Hutchings and Fraser 2008; Blanchet et al. 2008). Therefore, stocking with captive-bred
individuals should be expected to decrease diversity in natural populations (Eldridge et al. 2009;
Gossieaux et al. 2019), which has been a conservation and management conundrum for decades
(Hansen 2002; Hansen et al. 2009). Our results clearly contrast with this prediction, as we found a
positive relation between genetic diversity and captive-bred ancestry. Marie et al. (2010, 2012)
and Almodóvar et al. (2006) show that this is due to differences in origins and genetic drift
processes between captive-bred and wild populations, thus introducing new alleles into wild
populations. This result must thus be taken very cautiously because even if it increases genetic
diversity locally, stocking could overall decrease the species‘ genetic diversity and effective
population sizes (Almodóvar et al. 2006; Gossieaux et al. 2019).
When considering populations composed of individuals mostly assigned to the native cluster (i.e.
the supposedly native individuals), we found an overall downstream increase in genetic diversity
(DIGD), meeting the theoretical expectation (Ritland 1989; Morrissey and de Kerckhove 2009;
Paz-Vinas and Blanchet 2015; Paz-Vinas et al. 2015). Conversely, in populations composed of
individuals mostly assigned to the captive-bred cluster, genetic diversity significantly increased
upstream. Therefore, as captive-breeding ancestry is mostly present upstream, and harbors higher
levels of diversity, its observed effect on spatial patterns of diversity could simply be the result of
the distribution of strains. Moreover, the theoretical differentiation pattern (upstream increase in
differentiation) was weaker when considering individuals of high captive-bred ancestry. We
therefore suggest that stocking has homogenized genotypes among locations in which captive-bred
ancestry is present, and that genetic differentiation between captive-bred and introgressed
populations is lower than between wild populations (Hansen et al. 2006; Eldridge and Naish 2007;
Halbisen and Wilson 2009; Marie et al. 2010). Captive-bred strains are thus now maintained in
upper reaches in which naturally, populations would have diverged (Labonne et al. 2008; Paz-
Vinas and Blanchet 2015). Moreover, in Saint-Pé et al, (2018), we found that captive-bred
individuals tended to exhibit higher propensity for movement, moved on longer distances, and
moved preferentially from and towards tributaries (in which they are mostly present). This could
122
thus explain why these upstream populations of captive-bred fish and/or descendants are less
differentiated than the theoretical expectation (when comparing to the ―Fst-distance to source‖
slope of wild individuals).
Conclusions and implications
This study, along with others before it, confirms that the effects of stocking on spatial population
patterns can be substantial, and could highly affect conservation and management decisions. In
most of Europe, stocking has been highly underestimated both in terms of timescale and quantities
(Horreo and Garcia-Vazquez 2011; Miró and Ventura 2013). Its effects are thus still to be
assessed, as it is a major threat to species‘ genetic integrity, but also potentially to ecosystems.
Indeed, despite its potential direct effects on fitness (Hansen et al. 2009; Lamaze et al. 2012;
Morissette et al. 2018), its effects on spatial patterns of genetic diversity and differentiation also
have strong potential implications, and exist even in cases in which it was stopped many
generations ago (Hansen et al. 2009; Marie et al. 2010). Modifying spatial or temporal structure of
populations can for instance profoundly affect local food webs, with important subsequent
ecological consequences through community structure of invertebrate prey and primary
productivity for instance (Harmon et al. 2009; Raffard et al. 2017), but also through non-trophic
effects (Whitham et al. 2003; Matthews et al. 2011). Finally, the potential disappearance of
conservation units of high interest for both representation and persistence of diversity is another
conundrum posed by stocking (Antunes et al. 2001; Moritz 2002; Ferguson 2004). Indeed, as these
patterns of genetic diversity and differentiation can be ―context-dependent‖, we suggest that they
form very unique populations, which with stocking, are potentially threatened through changes in
these patterns.
Authors' contributions
KSP, SB, NP and LT designed the experiment and coordinated the study; KSP, LT, NP, and SB
conducted sampling; KSP and SB carried out the experimental lab work; KSP ran the statistical
analyses; KSP, LT, NP, and SB interpreted the data; KSP and SB wrote the first draft of the
manuscript; LT and NP, read, commented and corrected the initial draft, and all authors gave final
approval for publication.
123
Acknowledgments
We warmly thank the AAPPMAs of Luchon (J. Lérimé, D. Estrade and colleagues) and of
Saugues (J.C. Pulveric and colleagues), ECOGEA (P. Baran and colleagues), The Fédération
départementale pour le pêche et la protection des Milieux Aquatiques of Ariège (Laurent
Garmendia and collegues), of Haute-Garonne (Olivier Plasseraud and G. Durbe), of Alpes-
Maritimes and of Haute-Loire, and members of EDF (A. Barillier, V. Gouraud and colleagues) for
field sampling support. This work was undertaken at SETE, which forms part of the ―Laboratoire
d‘Excellence‖ (LABEX) entitled TULIP (ANR-10-LABX-41).
Funding
This work was funded by the French Agency for Biodiversity (n° 136406), Electricité De France
(n°8619-5920011137). Two of the authors are attached to the funding body (Laurence Tissot,
EDF; Nicolas Poulet, AFB), and have contributed to the design of the study, the collection,
analysis, and interpretation of data, and in writing the manuscript.
Conflict of Interest Statement
We have no competing interests and we declare no conflict of interests.
References
Aarestrup K, Jepsen N, Koed A, Pedersen S (2005) Movement and mortality of stocked brown trout in a
stream. J Fish Biol 66:721–728. doi: 10.1111/j.0022-1112.2005.00634.x
Almodóvar A, Nicola GG, Ayllón D, Elvira B (2012) Global warming threatens the persistence of
Mediterranean brown trout. Glob Change Biol 18:1549–1560. doi: 10.1111/j.1365-
2486.2011.02608.x
Almodovar A, Nicola GG, Elvira B, Garcia-Marin JL (2006) Introgression variability among Iberian brown
trout Evolutionary Significant Units: the influence of local management and environmental
features. Freshw Biol 51:1175–1187. doi: 10.1111/j.1365-2427.2006.01556.x
Almodovar A, Suarez J, Nicola GG, Nuevo M (2001) Genetic introgression between wild and stocked
brown trout in the Douro River basin, Spain. J Fish Biol 59:68–74. doi: 10.1111/j.1095-
8649.2001.tb01379.x
Alp M, Keller I, Westram AM, Robinson CT (2012) How river structure and biological traits influence
gene flow: a population genetic study of two stream invertebrates with differing dispersal abilities:
Biological traits and gene flow in stream invertebrates. Freshw Biol 57:969–981. doi:
10.1111/j.1365-2427.2012.02758.x
Antunes A, Faria R, Weiss S, Alexandrino P (2001) Complex evolutionary history in the brown trout:
insights on the recognition of conservation units. Conserv Genet 2:337–347. doi:
10.1023/A:1012518112891
124
Araguas RM, Sanz N, Pla C, Garcia-Marin JL (2004) Breakdown of the brown trout evolutionary history
due to hybridization between native and cultivated fish. J Fish Biol 65:28–37. doi: 10.1111/j.0022-
1112.2004.00544.x
Araki H, Schmid C (2010) Is hatchery stocking a help or harm? Aquaculture 308:S2–S11. doi:
10.1016/j.aquaculture.2010.05.036
Aurelle D, Berrebi P (2001) Genetic structure of brown trout (Salmo trutta, L.) populations from south-
western France: data from mitochondrial control region variability. Mol Ecol 10:1551–1561. doi:
10.1046/j.1365-294X.2001.01293.x
Barbat-Leterrier A, Guyomard R, Krieg F (1989) Introgression between introduced domesticated strains
and mediterranean native populations of brown trout (Salmo trutta L.). Aquat Living Resour
2:215–223. doi: 10.1051/alr:1989026
Benda L, Poff NL, Miller D, et al (2004) The Network Dynamics Hypothesis: how channel networks
structure riverine habitats. BioScience 54:413. doi: 10.1641/0006-
3568(2004)054[0413:TNDHHC]2.0.CO;2
Bernatchez L (2001) The evolutionary history of brown trout (Salmo trutta L.) inferred from
phylogeographic, nested clade, and mismatch analyses of mitochondrial dna variation. Evolution
55:351. doi: 10.1554/0014-3820(2001)055[0351:TEHOBT]2.0.CO;2
Berrebi P (2015) Three brown trout Salmo trutta lineages in Corsica described through allozyme variation:
salmo trutta genetics in corsica. J Fish Biol 86:60–73. doi: 10.1111/jfb.12534
Berrebi P, Poteaux C, Fissier M, Cattaneo-Berrebi G (2000) Stocking impact and allozyme diversity in
brown trout from Mediterranean southern France. J Fish Biol 56:949–960. doi: 10.1111/j.1095-
8649.2000.tb00884.x
Blanchet S, Páez DJ, Bernatchez L, Dodson JJ (2008) An integrated comparison of captive-bred and wild
Atlantic salmon (Salmo salar): Implications for supportive breeding programs. Biol Conserv
141:1989–1999. doi: 10.1016/j.biocon.2008.05.014
Blanchet S, Rey O, Etienne R, et al (2010) Species-specific responses to landscape fragmentation:
implications for management strategies. Evol Appl 3:291–304. doi: 10.1111/j.1752-
4571.2009.00110.x
Bohling JH (2016) Strategies to address the conservation threats posed by hybridization and genetic
introgression. Biol Conserv 203:321–327. doi: 10.1016/j.biocon.2016.10.011
Borsuk ME, Reichert P, Peter A, et al (2006) Assessing the decline of brown trout (Salmo trutta) in Swiss
rivers using a Bayesian probability network. Ecol Model 192:224–244. doi:
10.1016/j.ecolmodel.2005.07.006
Burnside RJ, Collar NJ, Scotland KM, Dolman PM (2016) Survival rates of captive-bred Asian Houbara
Chlamydotis macqueenii in a hunted migratory population. Ibis 158:353–361. doi:
10.1111/ibi.12349
Butler JRA, Radford A, Riddington G, Laughton R (2009) Evaluating an ecosystem service provided by
Atlantic salmon, sea trout and other fish species in the River Spey, Scotland: The economic impact
of recreational rod fisheries. Fish Res 96:259–266. doi: 10.1016/j.fishres.2008.12.006
Cagigas ME, Vazquez E, Blanco G, Sanchez JA (1999) Genetic effects of introduced hatchery stocks on
indigenous brown trout (Salmo trutta L.) populations in Spain. Ecol Freshw Fish 8:141–150. doi:
10.1111/j.1600-0633.1999.tb00065.x
Campbell Grant EH, Lowe WH, Fagan WF (2007) Living in the branches: population dynamics and
ecological processes in dendritic networks. Ecol Lett 10:165–175. doi: 10.1111/j.1461-
0248.2006.01007.x
Carrara F, Rinaldo A, Giometto A, Altermatt F (2014) Complex interaction of dendritic connectivity and
hierarchical patch size on biodiversity in river-like landscapes. Am Nat 183:13–25. doi:
10.1086/674009
125
Caudron A, Champigneulle A, Guyomard R (2006) Assessment of restocking as a strategy for
rehabilitating a native population of brown trout Salmo trutta L. in a fast-flowing mountain stream
in the northern French Alps. J Fish Biol 69:127–139. doi: 10.1111/j.1095-8649.2006.01156.x
Caudron A, Champigneulle A, Vigier L, et al (2012) Early effects of the strategies of creating a genetic
refuge and direct translocation for conserving and restoring populations of native brown trout:
native trout conservation strategies. Freshw Biol 57:1702–1715. doi: 10.1111/j.1365-
2427.2012.02823.x
Charlesworth B, Nordborg M, Charlesworth D (1997) The effects of local selection, balanced
polymorphism and background selection on equilibrium patterns of genetic diversity in subdivided
populations. Genet Res 70:155–174
Chave J (2013) The problem of pattern and scale in ecology: what have we learned in 20 years? Ecol Lett
16:4–16. doi: 10.1111/ele.12048
Christie MR, Marine ML, Fox SE, et al (2016) A single generation of domestication heritably alters the
expression of hundreds of genes. Nat Commun 7:10676. doi: 10.1038/ncomms10676
Christie MR, Marine ML, French RA, et al (2012) Effective size of a wild salmonid population is greatly
reduced by hatchery supplementation. Heredity 109:254–260. doi: 10.1038/hdy.2012.39
Comte L, Buisson L, Daufresne M, Grenouillet G (2013) Climate-induced changes in the distribution of
freshwater fish: observed and predicted trends: Climate change and freshwater fish. Freshw Biol
58:625–639. doi: 10.1111/fwb.12081
Cyr F, Angers B (2011) Historical process lead to false genetic signal of current connectivity among
populations. Genetica 139:1417–1428. doi: 10.1007/s10709-012-9640-2
de Sostoa A, Lobon-Cervia J (1989) Observations on feeding relationships between fish predators and fish
assemblages in a Mediterranean stream. Regul Rivers Res Manag 4:157–163. doi:
10.1002/rrr.3450040208
Earl DA, vonHoldt BM (2012) STRUCTURE HARVESTER: a website and program for visualizing
STRUCTURE output and implementing the Evanno method. Conserv Genet Resour 4:359–361.
doi: 10.1007/s12686-011-9548-7
Eldridge WH, Myers JM, Naish KA (2009) Long-term changes in the fine-scale population structure of
coho salmon populations (Oncorhynchus kisutch) subject to extensive supportive breeding.
Heredity 103:299–309. doi: 10.1038/hdy.2009.69
Eldridge WH, Naish KA (2007) Long-term effects of translocation and release numbers on fine-scale
population structure among coho salmon (Oncorhynchus kisutch). Mol Ecol 16:2407–2421. doi:
10.1111/j.1365-294X.2007.03271.x
Elliott JM (1994) Quantitative ecology and the brown trout. Oxford University Press, Oxford ; New York
Ellstrand NC, Elam DR (1993) Population genetic consequences of small population size: implications for
plant conservation. Annu Rev Ecol Syst 24:217–242. doi: 10.1146/annurev.es.24.110193.001245
Estoup A, Largiadèr CR, Cornuet JM, et al (2000) Juxtaposed microsatellite systems as diagnostic markers
for admixture: an empirical evaluation with brown trout (Salmo trutta) as model organism. Mol
Ecol 9:1873–1886
Evanno G, Regnaut S, Goudet J (2005) Detecting the number of clusters of individuals using the software
structure: a simulation study. Mol Ecol 14:2611–2620. doi: 10.1111/j.1365-294X.2005.02553.x
Faulks LK, Gilligan DM, Beheregaray LB (2011) The role of anthropogenic vs. natural in-stream structures
in determining connectivity and genetic diversity in an endangered freshwater fish, Macquarie
perch (Macquaria australasica): anthropogenic vs. natural in-stream structures in M. australasica.
Evol Appl 4:589–601. doi: 10.1111/j.1752-4571.2011.00183.x
Ferguson A (2004) The importance of identifying conservation units: brown trout and pollan biodiversity in
Ireland. Biol Environ Proc R Ir Acad 104:33–41. doi: 10.3318/BIOE.2004.104.3.33
Finnegan AK, Stevens JR (2008) Assessing the long-term genetic impact of historical stocking events on
contemporary populations of Atlantic salmon, Salmo salar. Fish Manag Ecol 15:315–326. doi:
10.1111/j.1365-2400.2008.00616.x
126
Frankham R, Ballou JD, Briscoe DA (2010) Introduction to conservation genetics, 2nd ed. Cambridge
University Press, Cambridge, UK ; New York
Frankham R, Hemmer H, Ryder OA, et al (1986) Selection in captive populations. Zoo Biol 5:127–138.
doi: 10.1002/zoo.1430050207
Franklin IR, Frankham R (1998) How large must populations be to retain evolutionary potential? Anim
Conserv 1:69–70. doi: 10.1111/j.1469-1795.1998.tb00228.x
García‐Marín, J.L., Sanz, N. and Pla, C., 1999. Erosion of the native genetic resources of brown trout in
Spain. Ecology of Freshwater fish, 8(3), pp.151-158.
Gaston KJ (2000) Global patterns in biodiversity. Nature 405:220–227. doi: 10.1038/35012228
Gossieaux P, Bernatchez L, Sirois P, Garant D (2019) Impacts of stocking and its intensity on effective
population size in Brook Charr (Salvelinus fontinalis) populations. Conserv Genet. doi:
10.1007/s10592-019-01168-2
Grant EHC, Lynch HJ, Muneepeerakul R, et al (2012) Interbasin water transfer, riverine connectivity, and
spatial controls on fish biodiversity. PLoS ONE 7:e34170. doi: 10.1371/journal.pone.0034170
Halbisen MA, Wilson CC (2009) Variable Introgression from supplemental stocking in southern Ontario
populations of lake trout. Trans Am Fish Soc 138:699–719. doi: 10.1577/T07-135.1
Hansen MM (2002) Estimating the long-term effects of stocking domesticated trout into wild brown trout
(Salmo trutta) populations: an approach using microsatellite DNA analysis of historical and
contemporary samples. Mol Ecol 11:1003–1015. doi: 10.1046/j.1365-294X.2002.01495.x
Hansen MM, Bekkevold D, Jensen LF, et al (2006) Genetic restoration of a stocked brown trout Salmo
trutta population using microsatellite DNA analysis of historical and contemporary samples:
Genetic restoration of brown trout populations. J Appl Ecol 43:669–679. doi: 10.1111/j.1365-
2664.2006.01185.x
Hansen MM, Fraser DJ, Meier K, Mensberg K-LD (2009) Sixty years of anthropogenic pressure: a spatio-
temporal genetic analysis of brown trout populations subject to stocking and population declines.
Mol Ecol 18:2549–2562. doi: 10.1111/j.1365-294X.2009.04198.x
Hansen MM, Hynes RA, Loeschcke V, Rasmussen G (1995) Assessment of the stocked or wild origin of
anadromous brown trout (Salmo trutta L.) in a Danish river system, using mitochondrial DNA
RFLP analysis. Mol Ecol 4:189–198. doi: 10.1111/j.1365-294X.1995.tb00208.x
Hansen MM, Nielsen EE, Bekkevold D, Mensberg K-LD (2001) Admixture analysis and stocking impact
assessment in brown trout (Salmo trutta), estimated with incomplete baseline data. Can J Fish
Aquat Sci 58:1853–1860. doi: 10.1139/cjfas-58-9-1853
Harbicht AB, Alshamlih M, Wilson CC, Fraser DJ (2014) Anthropogenic and habitat correlates of
hybridization between hatchery and wild brook trout. Can J Fish Aquat Sci 71:688–697. doi:
10.1139/cjfas-2013-0460
Harmon LJ, Matthews B, Des Roches S, et al (2009) Evolutionary diversification in stickleback affects
ecosystem functioning. Nature 458:1167–1170. doi: 10.1038/nature07974
Heggenes J, Roed KH, Hoyheim B, Rosef L (2002) Microsatellite diversity assessment of brown trout
(Salmo trutta) population structure indicate limited genetic impact of stocking in a Norwegian
alpine lake. Ecol Freshw Fish 11:93–100. doi: 10.1034/j.1600-0633.2002.00009.x
Horreo JL, Garcia-Vazquez E (2011) Foreign brown trout in protected landscapes as a consequence of
connectivity: brown trout, protected landscapes and connectivity. Fish Manag Ecol 18:431–436.
doi: 10.1111/j.1365-2400.2011.00803.x
Hutchings JA, Fraser DJ (2008) The nature of fisheries- and farming-induced evolution. Mol Ecol 17:294–
313. doi: 10.1111/j.1365-294X.2007.03485.x
Jakobsson M, Rosenberg NA (2007) CLUMPP: a cluster matching and permutation program for dealing
with label switching and multimodality in analysis of population structure. Bioinformatics
23:1801–1806. doi: 10.1093/bioinformatics/btm233
Jombart T (2008) adegenet: a R package for the multivariate analysis of genetic markers. Bioinformatics
24:1403–1405. doi: 10.1093/bioinformatics/btn129
127
Kazyak DC, Rash J, Lubinski BA, King TL (2018) Assessing the impact of stocking northern-origin
hatchery brook trout on the genetics of wild populations in North Carolina. Conserv Genet 19:207–
219. doi: 10.1007/s10592-017-1037-4
Keller I, Largiadèr CR (2003) Recent habitat fragmentation caused by major roads leads to reduction of
gene flow and loss of genetic variability in ground beetles. Proc R Soc Lond B Biol Sci 270:417–
423. doi: 10.1098/rspb.2002.2247
Kikuchi S, Suzuki W, Sashimura N (2011) Gene flow in an endangered willow Salix hukaoana (Salicaceae)
in natural and fragmented riparian landscapes. Conserv Genet 12:79–89. doi: 10.1007/s10592-009-
9992-z
Krieg F, Guyomard R (1985) Population genetics of French brown trout (Salmo trutta L): large
geographical differentiation of wild populations and high similarity of domesticated stocks. Genet
Sel Evol 17:225. doi: 10.1186/1297-9686-17-2-225
Labonne J, Ravigné V, Parisi B, Gaucherel C (2008) Linking dendritic network structures to population
demogenetics: the downside of connectivity. Oikos 117:1479–1490. doi: 10.1111/j.0030-
1299.2008.16976.x
Laikre L, Antunes A, Concerted Action on Identification M and E ofGenetic R in the BT (Salmo trutta)
(projekt) (1999) Conservation genetic management of brown trout (Salmo trutta) in Europe.
Danmarks Fiskeriundersøgelser, Afd. for Ferskvandsfiskeri, Silkeborg
Lamaze FC, Sauvage C, Marie A, et al (2012) Dynamics of introgressive hybridization assessed by SNP
population genomics of coding genes in stocked brook charr (Salvelinus fontinalis): SNP
population genomics in Brook charr. Mol Ecol 21:2877–2895. doi: 10.1111/j.1365-
294X.2012.05579.x
Largiader CR, Scholl A (1996) Genetic introgression between native and introduced brown trout Salmo
trutta L. populations in the Rhone River basin. Mol Ecol 5:417–426. doi: 10.1046/j.1365-
294X.1996.00099.x
Largiadèr CR, Scholl A, Guyomard R (1996) The role of natural and artificial propagation on the genetic
diversity of brown trout (Salmo trutta L.) of the upper Rhône drainage. In: Kirchhofer A, Hefti D
(eds) Conservation of Endangered Freshwater Fish in Europe. Birkhäuser Basel, Basel, pp 181–197
Leitwein M, Guinand B, Pouzadoux J, et al (2017) A dense brown trout (Salmo trutta) linkage map reveals
recent chromosomal rearrangements in the Salmo genus and the impact of selection on linked
neutral diversity. G3amp58 GenesGenomesGenetics 7:1365–1376. doi: 10.1534/g3.116.038497
Létourneau J, Ferchaud A-L, Le Luyer J, et al (2018) Predicting the genetic impact of stocking in Brook
Charr (Salvelinus fontinalis) by combining RAD sequencing and modeling of explanatory
variables. Evol Appl 11:577–592. doi: 10.1111/eva.12566
Lobón-Cervía J, Sanz N (eds) (2017) Brown trout: biology, ecology and management. Wiley, Hoboken, NJ
Love Stowell SM, Kennedy CM, Beals SC, et al (2015) The genetic legacy of more than a century of
stocking trout: a case study in Rocky Mountain National Park, Colorado, USA. Can J Fish Aquat
Sci 72:1565–1574. doi: 10.1139/cjfas-2015-0009
Manel S, Schwartz MK, Luikart G, Taberlet P (2003) Landscape genetics: combining landscape ecology
and population genetics. Trends Ecol Evol 18:189–197. doi: 10.1016/S0169-5347(03)00008-9
Marie AD, Bernatchez L, Garant D (2010) Loss of genetic integrity correlates with stocking intensity in
brook charr (Salvelinus fontinalis): impacts of stocking on Salvelinus fontinalis. Mol Ecol 19:2025–
2037. doi: 10.1111/j.1365-294X.2010.04628.x
Marie AD, Bernatchez L, Garant D (2012) Environmental factors correlate with hybridization in stocked
brook charr (Salvelinus fontinalis). Can J Fish Aquat Sci 69:884–893. doi: 10.1139/f2012-027
Martinez P, Arias J, Castro J, Sanchez L (1993) Differential stocking incidence in brown trout (Salmo
trutta) populations from Northwestern Spain. Aquaculture 114:203–216. doi: 10.1016/0044-
8486(93)90296-B
128
Matthews B, Narwani A, Hausch S, et al (2011) Toward an integration of evolutionary biology and
ecosystem science: Integration of evolutionary biology and ecosystem science. Ecol Lett 14:690–
701. doi: 10.1111/j.1461-0248.2011.01627.x
Mezzera, M. and Largiader, C.R., 2001. Evidence for selective angling of introduced trout and their hybrids
in a stocked brown trout population. Journal of Fish Biology, 59(2), pp.287-301.
Mills D (1989) Conservation and management of brown trout, Salmo trutta, in Scotland: a historical review
and the future. Freshw Biol 21:87–98. doi: 10.1111/j.1365-2427.1989.tb01350.x
Mimura M, Yahara T, Faith DP, et al (2017) Understanding and monitoring the consequences of human
impacts on intraspecific variation. Evol Appl 10:121–139. doi: 10.1111/eva.12436
Miraldo A, Li S, Borregaard MK, et al (2016) An Anthropocene map of genetic diversity. Science
353:1532–1535. doi: 10.1126/science.aaf4381
Miró A, Ventura M (2013) Historical use, fishing management and lake characteristics explain the presence
of non-native trout in Pyrenean lakes: implications for conservation. Biol Conserv 167:17–24. doi:
10.1016/j.biocon.2013.07.016
Moran P, Perez J, Dumas J, et al (2005) Stocking-related patterns of genetic variation at enzymatic loci in
south European Atlantic salmon populations. J Fish Biol 67:185–199. doi: 10.1111/j.0022-
1112.2005.00847.x
Morissette O, Sirois P, Lester NP, et al (2018) Supplementation stocking of Lake Trout (Salvelinus
namaycush) in small boreal lakes: ecotypes influence on growth and condition. PLOS ONE
13:e0200599. doi: 10.1371/journal.pone.0200599
Moritz C (2002) Strategies to protect biological diversity and the evolutionary processes that sustain it. Syst
Biol 51:238–254. doi: 10.1080/10635150252899752
Morrissey MB, de Kerckhove DT (2009) The maintenance of genetic variation due to asymmetric gene
flow in dendritic metapopulations. Am Nat 174:875–889. doi: 10.1086/648311
Paz-Vinas I, Blanchet S (2015) Dendritic connectivity shapes spatial patterns of genetic diversity: a
simulation-based study. J Evol Biol 28:986–994. doi: 10.1111/jeb.12626
Paz-Vinas I, Loot G, Hermoso V, et al (2018) Systematic conservation planning for intraspecific genetic
diversity. Proc R Soc B Biol Sci 285:20172746. doi: 10.1098/rspb.2017.2746
Paz-Vinas I, Loot G, Stevens VM, Blanchet S (2015) Evolutionary processes driving spatial patterns of
intraspecific genetic diversity in river ecosystems. Mol Ecol 24:4586–4604. doi:
10.1111/mec.13345
Pedersen L-F, Koed A, Malte H (2008) Swimming performance of wild and F1-hatchery-reared Atlantic
salmon (Salmo salar) and brown trout (Salmo trutta) smolts. Ecol Freshw Fish 17:425–431. doi:
10.1111/j.1600-0633.2008.00293.x
Perrier C, Baglinière J-L, Evanno G (2013a) Understanding admixture patterns in supplemented
populations: a case study combining molecular analyses and temporally explicit simulations in
Atlantic salmon. Evol Appl 6:218–230. doi: 10.1111/j.1752-4571.2012.00280.x
Perrier C, Guyomard R, Bagliniere J-L, et al (2013b) Changes in the genetic structure of Atlantic salmon
populations over four decades reveal substantial impacts of stocking and potential resiliency. Ecol
Evol 3:2334–2349. doi: 10.1002/ece3.629
Peterson EE, Hoef JMV (2014) STARS : An ArcGIS toolset used to calculate the spatial information
needed to fit spatial statistical models to stream network data. J Stat Softw 56:. doi:
10.18637/jss.v056.i02
Post JR (2013) Resilient recreational fisheries or prone to collapse? A decade of research on the science and
management of recreational fisheries. Fish Manag Ecol 20:99–110. doi: 10.1111/fme.12008
Poteaux C (1999) Microsatellite polymorphism and genetic impact of restocking in Mediterranean brown
trout (Salmo trutta L.). Heredity 82:645. doi: 10.1046/j.1365-2540.1999.00519.x
Pritchard JK, Stephens M, Donnelly P (2000) Inference of population structure using multilocus genotype
data. Genetics 155:945–959
129
Prunier JG, Dubut V, Loot G, et al (2018) The relative contribution of river network structure and
anthropogenic stressors to spatial patterns of genetic diversity in two freshwater fishes: A multiple-
stressors approach. Freshw Biol 63:6–21. doi: 10.1111/fwb.13034
Raeymaekers JAM, Maes GE, Geldof S, et al (2008) Modeling genetic connectivity in sticklebacks as a
guideline for river restoration. Evol Appl 1:475–488. doi: 10.1111/j.1752-4571.2008.00019.x
Raffard A, Lecerf A, Cote J, et al (2017) The functional syndrome: linking individual trait variability to
ecosystem functioning. Proc R Soc B Biol Sci 284:20171893. doi: 10.1098/rspb.2017.1893
Ritland K (1989) Genetic differentiation, diversity, and inbreeding in the mountain monkeyflower
(Mimulus caespitosus) of the Washington Cascades. Can J Bot 67:2017–2024. doi: 10.1139/b89-
255
Rosenberg NA (2003) distruct: a program for the graphical display of population structure: program note.
Mol Ecol Notes 4:137–138. doi: 10.1046/j.1471-8286.2003.00566.x
Saint-Pé K, Blanchet S, Tissot L, et al (2018) Genetic admixture between captive-bred and wild individuals
affects patterns of dispersal in a brown trout (Salmo trutta) population. Conserv Genet 19:1269–
1279. doi: 10.1007/s10592-018-1095-2
Smith SM, Maughan PJ (2015) SNP Genotyping Using KASPar assays. In: Batley J (ed) Plant genotyping.
Springer New York, New York, NY, pp 243–256
Splendiani A, Ruggeri P, Giovannotti M, Caputo Barucchi V (2013) Role of environmental factors in the
spread of domestic trout in Mediterranean streams. Freshw Biol 58:2089–2101. doi:
10.1111/fwb.12193
Swatdipong A, Vasemägi A, Niva T, et al (2010) High level of population genetic structuring in lake-run
brown trout, Salmo trutta, of the Inari Basin, northern Finland. J Fish Biol 77:2048–2071. doi:
10.1111/j.1095-8649.2010.02784.x
Torterotot J-B, Perrier C, Bergeron NE, Bernatchez L (2014) Influence of forest road culverts and
waterfalls on the fine-scale distribution of Brook trout genetic diversity in a Boreal Watershed.
Trans Am Fish Soc 143:1577–1591. doi: 10.1080/00028487.2014.952449
Vähä J-P, Primmer CR (2005) Efficiency of model-based Bayesian methods for detecting hybrid
individuals under different hybridization scenarios and with different numbers of loci: genetic
detection of hybrids. Mol Ecol 15:63–72. doi: 10.1111/j.1365-294X.2005.02773.x
Valiquette E, Perrier C, Thibault I, Bernatchez L (2014) Loss of genetic integrity in wild lake trout
populations following stocking: insights from an exhaustive study of 72 lakes from Québec,
Canada. Evol Appl 7:625–644. doi: 10.1111/eva.12160
Vannote RL, Minshall GW, Cummins KW, et al (1980) The river continuum concept. Can J Fish Aquat Sci
37:130–137. doi: 10.1139/f80-017
Vasemägi A, Gross R, Paaver T, et al (2005) Extensive immigration from compensatory hatchery releases
into wild Atlantic salmon population in the Baltic sea: spatio-temporal analysis over 18 years.
Heredity 95:76–83. doi: 10.1038/sj.hdy.6800693
Vøllestad LA, Serbezov D, Bass A, et al (2012) Small-scale dispersal and population structure in stream-
living brown trout (Salmo trutta) inferred by mark–recapture, pedigree reconstruction, and
population genetics. Can J Fish Aquat Sci 69:1513–1524. doi: 10.1139/f2012-073
Weber ED, Fausch KD (2003) Interactions between hatchery and wild salmonids in streams: differences in
biology and evidence for competition. Can J Fish Aquat Sci 60:1018–1036. doi: 10.1139/f03-087
White SL, Miller WL, Dowell SA, et al (2018) Limited hatchery introgression into wild brook trout
(Salvelinus fontinalis) populations despite reoccurring stocking. Evol Appl 11:1567–1581. doi:
10.1111/eva.12646
Whitham TG, Young WP, Martinsen GD, et al (2003) Community and ecosystem genetics: a consequence
of the extended phenotype. Ecology 84:559–573. doi: 10.1890/0012-
9658(2003)084[0559:CAEGAC]2.0.CO;2
130
Supplementary material for Chapter III.
Appendix S3.1: Fish sampled; numbers and sizes
Basin N Mean body length (mm) Mean Fst Mean He Mean Fis
Aude 453 143(±37) 0.114(±0.112) 0.162(±0.089) 0.032(±0.086) One 495 136(±38) 0.063(±0.024) 0.139(±0.022) 0.003(±0.034) Roya 413 167(±23) 0.091(±0.043) 0.305(±0.064) 0.024(±0.031) Seuge 377 134(±34) 0.069(±0.021) 0.091(±0.019) -0.036(±0.112)
Basin Site N Mean body length (mm)(±SE)
Uniqueness (mean pairwise Fst) He Fis X Y
AU
DE
AGUSou 27 157(±42) 0.071 0.196 0.058 627088.1 6185199.75
AIGPou 30 139(±33) 0.096 0.072 0.01 637243.95 6183735
AIGSou 30 143(±38) 0.071 0.181 0.018 636876.2 6180792.25
ARTLau 30 138(±38) 0.113 0.069 0.018 621320.8 6176817
AUDCar 30 166(±34) 0.093 0.074 0.027 627921.65 6180545.5
AUDFar 30 165(±33) 0.068 0.139 -0.028 625004.1 6183550.5
AUDFou 30 135(±25) 0.073 0.126 0.059 629048.35 6186449
AUDNen 30 123(±54) 0.081 0.098 0.001 635955.25 6186755.75
AUDPuy 30 121(±31) 0.085 0.288 -0.092 628263.15 6173847.75
AUDSer 30 156(±22) 0.067 0.144 0.004 632930.0927 6186461.501
BAIPas 14 153(±28) 0.534 0.177 0.090 638128.2 6181735.75
BRUMij 30 127(±26) 0.103 0.066 0.059 621999.45 6181223.75
BRUUss 30 151(±21) 0.082 0.103 0.047 625055 6182171.25
CAMSau 14 118(±46) 0.085 0.156 -0.129 624813.2 6184020.75
QUEMas 29 131(±36) 0.067 0.262 0.262 625591 6178665.75
QUERia 30 158(±37) 0.063 0.213 0.137 624547.7 6181059.25
ROQSau 9 154(±25) 0.191 0.388 -0.004 635481.2 6182155.75
ON
E
COULar 30 111(±26) 0.135 0.091 0.028 494582.72 6191615.94
LABCas 17 112(±23) 0.061 0.14 0.042 497950.61 6192263.67
NGAJur 30 152(±40) 0.087 0.151 -0.004 494142.91 6194213.37
NGAMar 30 135(±18) 0.097 0.161 -0.016 494232.47 6193507.03
NGAVga 30 127(±35) 0.062 0.166 -0.053 496537.4 6193069.2
NOOAst 30 135(±26) 0.051 0.12 -0.03 495260.7 6188546.68
NOOCas 30 144(±18) 0.045 0.129 0.065 497578.93 6192335.96
NOOEsp 30 129(±38) 0.055 0.118 -0.023 495647.88 6189592.87
NOOLac 30 109(±26) 0.054 0.115 -0.011 495122.63 6187988.33
NOOSav 30 111(±63) 0.044 0.131 -0.002 499144.09 6192586.61
NOOTre 32 157(±33) 0.049 0.128 -0.014 500610.5 6192174.74
NOOVoo 28 138(±30) 0.043 0.138 0.043 496043.43 6191942.32
NOUBoo 28 173(±43) 0.064 0.182 0.07 495090.59 6198734.52
NOUCir 30 134(±42) 0.067 0.156 0 496859.39 6197875.96
131
NOUMay 30 155(±34) 0.053 0.152 -0.001 498770.59 6196493.34
NOUSpo 30 140(±38) 0.064 0.134 -0.02 499342.58 6194843.9
ONEBag 30 147(±28) 0.042 0.146 -0.015 501655.8 6191325.5 SE
UG
E
BuiPie 6 96(±27) 0.1 0.041 0.044 736154.95 6421760.5
CLATou 9 105(±55) 0.07 0.08 -0.006 740628.2 6424059.5
LAVRiv 15 138(±51) 0.064 0.105 -0.018 738719.1508 6430339.667
LAVSuc 30 122(±31) 0.053 0.102 -0.097 740929.25 6428709.75
PONAmo 30 127(±26) 0.056 0.091 -0.01 736948.8429 6427435.268
PONBom 30 133(±32) 0.084 0.072 -0.08 735932.9 6423741.5
PONCha 30 148(±29) 0.048 0.108 -0.014 742060.2 6428955.5
PONTet 11 114(±28) 0.112 0.061 -0.269 734965.3332 6421406.405
PONTis 30 129(±18) 0.045 0.11 -0.006 739554.4898 6428659.981
SERSau 6 143(±28) 0.057 0.098 0.125 737591.5467 6428895.11
SEUCch 30 110(±21) 0.098 0.078 -0.162 736912.1 6419850.75
SEUCou 30 134(±27) 0.059 0.102 -0.081 742646.6581 6424696.732
SEUCro 30 165(±33) 0.08 0.1 0.151 742507.05 6426326.75
SEUFag 30 126(±32) 0.078 0.097 -0.21 737803.35 6423289.5
SEURod 30 157(±38) 0.05 0.102 0.059 743269.65 6430201.25
SEUSau 30 138(±19) 0.055 0.101 0.006 742641.15 6428719.25
RO
YA
BENCas 8 158(±33) 0.048 0.37 0.043 1064681.159 6330776.294
BIECas 30 157(±47) 0.076 0.398 0.062 1061599.65 6342372.25
BIEMai 30 172(±33) 0.153 0.346 -0.029 1065152.9 6339425
BIEMin 30 171(±36) 0.156 0.34 0.024 1060994.8 6340253.25
CAIGaf 30 170(±48) 0.067 0.293 0.01 1061323.924 6332296.13
LEVBri 30 140(±27) 0.057 0.352 -0.028 1070230.357 6340355.202
LEVTen 30 153(±30) 0.069 0.386 0.004 1067621.35 6339593.25
MAGCem 30 176(±32) 0.075 0.272 0.041 1061971.15 6327821.25
REFRic 30 123(±23) 0.133 0.363 -0.012 1068559.35 6343613.75
ROYAmb 30 171(±52) 0.083 0.243 0.041 1018145.09 1901873.82
ROYBre 30 193(±63) 0.085 0.238 0.012 1016478.36 1894038.21
ROYChi 24 196(±24) 0.081 0.244 -0.005 1064068.351 6329711.933
ROYDal 30 143(±27) 0.06 0.379 -0.002 1067460.706 6338535.476
ROYEvc 30 196(±59) 0.085 0.237 0.031 1016219.44 1895386.21
ROYFon 30 199(±65) 0.087 0.229 0.054 1018390.04 1902415.5
ROYGia 30 162(±47) 0.092 0.22 0.072 1016240.41 1897091.32
ROYOrt 30 131(±18) 0.229 0.267 -0.017 1065392.1 6346381.5
ROYPie 21 207(±31) 0.085 0.235 0.029 1062946.375 6322646.07
ROYSca 30 172(±47) 0.053 0.353 0.048 1065916.912 6335228.32
ROYTen 30 141(±27) 0.065 0.384 0.07 1068075.713 6341877.007
ROYVei 30 182(±40) 0.08 0.249 0.046 1016385.91 1897993.54
132
Appendix S3.2: Summary of individual proportion of captive-bred ancestry for each basin (in %)
Min Max Median Mean Sd
AUDE 1.5 98.4 7.1 16.3 24.3
ONE 1.6 94.4 8.3 22.6 30.9
SEUGE 0.5 33.6 1.6 3.7 8.0
ROYA 25.5 90.7 66.7 59.1 22.9
133
Appendix S3.3: Model selection for explaining spatial patterns of admixture (A), and spatial
patterns of genetic diversity (B) and differentiation (C)
(A)
Introg (Int) BV dist_drn BV:dist_drn df logLik AICc delta weight
4 -1.98 + 0.42
6.00 63.05 -112.70 0.00 0.87 8 -1.84 + 0.07 + 9.00 64.84 -108.60 4.10 0.11 2 -1.70 +
5.00 57.90 -104.80 7.89 0.02
3 -1.00
0.53
3.00 39.09 -71.80 40.92 0.00 1 -0.94 2.00 32.56 -60.90 51.81 0.00
134
(B) He (Int) BV dist_drn Introg BV:dist_drn BV:Introg dist_drn:Introg df logLik AICc delta weight
22.0 -2.2 +
4.2
+
9.0 155.9 -290.6 0.0 0.4
24.0 -2.1 + -6.82E-06 4.2
+
10.0 157.0 -290.3 0.4 0.3
56.0 -2.2 + -2.80E-06 4.4
+ -1.47E-05 11.0 157.6 -288.6 2.1 0.1
32.0 -2.3 + 4.32E-06 4.2 + +
13.0 159.9 -287.2 3.5 0.1
64.0 -2.3 + 7.78E-06 4.5 + + -2.56E-05 14.0 161.0 -286.2 4.5 0.0
6.0 -1.8 +
1.0
6.0 123.7 -234.0 56.6 0.0
8.0 -1.8 + -1.79E-06 1.0
7.0 123.7 -231.6 59.0 0.0
40.0 -1.7 + -1.02E-05 0.8
1.81E-05 8.0 124.5 -230.6 60.0 0.0
16.0 -1.9 + 7.40E-06 1.1 +
10.0 125.8 -227.8 62.9 0.0
48.0 -1.9 + 4.49E-06 0.7 +
2.89E-05 11.0 126.5 -226.3 64.3 0.0
4.0 -2.0 + 1.41E-05
6.0 112.0 -210.7 79.9 0.0
12.0 -1.9 + 8.67E-06
+
9.0 114.4 -207.8 82.8 0.0
2.0 -1.7 +
5.0 108.8 -206.7 83.9 0.0
5.0 -2.0
1.6
3.0 103.8 -201.2 89.5 0.0
7.0 -2.0
8.54E-06 1.4
4.0 104.7 -200.7 90.0 0.0
39.0 -1.9
-1.89E-06 1.1
2.39E-05 5.0 105.6 -200.3 90.3 0.0
3.0 -1.9
3.47E-05
3.0 77.5 -148.6 142.1 0.0
1.0 -1.5 2.0 69.6 -135.0 155.6 0.0
135
(C) Fst (Int) BV dist_drn Introg BV:dist_drn BV:Introg dist_drn:Introg df logLik AICc delta weight
3.00 0.05
2.02E-06
3.00 146.53 -286.70 0.00 0.35
7.00 0.05
1.68E-06 0.02
4.00 147.37 -286.10 0.59 0.26
39.00 0.06
1.29E-06 0.00
1.13E-06 5.00 147.67 -284.40 2.31 0.11
64.00 0.09 + -1.34E-06 -0.02 + + 1.13E-05 14.00 159.76 -283.70 2.95 0.08
48.00 0.09 + -7.76E-07 -0.06 +
8.30E-06 11.00 154.80 -283.00 3.73 0.06
56.00 0.07 + 1.66E-07 0.05
+ 6.37E-06 11.00 154.71 -282.80 3.90 0.05
8.00 0.07 + 1.25E-06 0.03
7.00 148.68 -281.50 5.18 0.03
4.00 0.06 + 1.67E-06
6.00 147.39 -281.40 5.27 0.03
40.00 0.08 + 5.14E-07 0.00
1.91E-06 8.00 149.47 -280.50 6.16 0.02
6.00 0.08 +
0.05
6.00 146.26 -279.20 7.54 0.01
24.00 0.05 + 1.47E-06 0.14
+
10.00 150.76 -277.70 8.96 0.00
5.00 0.07
0.04
3.00 142.02 -277.70 9.02 0.00
12.00 0.08 + 3.56E-07
+
9.00 148.73 -276.40 10.29 0.00
16.00 0.08 + 2.78E-07 0.03 +
10.00 149.96 -276.10 10.57 0.00
22.00 0.07 +
0.15
+
9.00 148.24 -275.40 11.28 0.00
2.00 0.09 +
5.00 142.21 -273.50 13.23 0.00
32.00 0.07 + -8.82E-09 0.15 + +
13.00 152.60 -272.60 14.12 0.00
1.00 0.08 2.00 136.89 -269.60 17.10 0.00
136
Appendix S4: Plot of expected heterozygosity (He) against riparian distance to the river mouth,
depending on the river basin as an interaction term
137
Appendix S3.5: Plot of genetic differentiation (Fst) against riparian distance to the river mouth,
depending on the river basin as an interaction term
138
Chapter IV. Genetic admixture between captive-bred and
wild individuals affects patterns of dispersal in a brown trout
(Salmo trutta) population
Keoni Saint-Pé 1*
, Simon Blanchet1,2*
, Laurence Tissot4, Nicolas Poulet
5, Olivier Plasseraud
6,
Géraldine Loot2,3
, Charlotte Veyssière1 and Jérôme G. Prunier
1
1 Centre National de la Recherche Scientifique (CNRS), Université Paul Sabatier (UPS); Station
d‘Ecologie Théorique et Expérimentale, UMR 5321, F-09200 Moulis, France
2 CNRS, UPS, École Nationale de Formation Agronomique (ENFA) ; UMR 5174 EDB
(Laboratoire Évolution & Diversité Biologique), 118 route de Narbonne, F-31062 Toulouse cedex
4, France
3 Université de Toulouse, UPS, UMR 5174 (EDB), 118 route de Narbonne, F-31062 Toulouse
cedex 4, France
4 EDF Recherche et Développement, Laboratoire National d‘Hydraulique et Environnement, 6
quai Watier, 78401 Chatou Cedex, France
5 Pôle écohydraulique AFB-IMT, allée du Pr Camille Soula, 31400 Toulouse France
6 Fédération Départementale pour la Pêche et la Protection des Milieux Aquatiques de Haute
Garonne, 395 route de Saint-Simon, 31100 Toulouse France
Key words: Genetic admixture, dispersal, brown trout, microsatellites, individual assignment,
captive breeding, stocking, hatchery.
Manuscript published in Conservation Genetics, Volume 19, Issue 5, pp 1269–1279
https://doi.org/10.1007/s10592-018-1095-2
139
Summary
1. Genetic admixture between captive-bred and wild individuals has been demonstrated to affect
many individual traits, although little is known about its potential influence on dispersal, an
important trait governing the eco-evolutionary dynamics of populations.
2. Here, we quantified and described the spatial distribution of genetic admixture in a brown trout
(Salmo trutta) population from a small watershed that was stocked until 1999, and then tested
whether or not individual dispersal parameters were related to admixture between wild and
captive-bred fish.
3. We genotyped 715 fish at 17 microsatellite loci sampled from both the mainstream and all
populated tributaries, as well as 48 fish from the hatchery used to stock the study area. First,
we used Bayesian clustering to infer local genetic structure and to quantify genetic admixture.
We inferred first generation migrants to identify dispersal events and test which features
(genetic admixture, sex and body length) affected dispersal parameters (i.e. probability to
disperse, distance of dispersal and direction of the dispersal event).
4. We identified two genetic clusters in the river basin, corresponding to wild fish on the one hand
and to fish derived from the captive strain on the other hand, allowing us to define an
individual gradient of admixture. Individuals with a strong assignment to the captive strain
occurred almost exclusively in some tributaries, and were more likely to disperse towards a
tributary than towards a site of the mainstream. Furthermore, dispersal probability increased as
the probability of assignment to the captive strain increased, and individuals with an
intermediate level of admixture exhibited the lowest dispersal distances.
5. These findings show that various dispersal parameters may be biased by admixture with
captive-bred genotypes, and that management policies should take into account the differential
spread of captive-bred individuals in wild populations.
140
Introduction
Captive breeding programs are a management practice that is commonly used to sustain
endangered species and populations (Seddon et al. 2007; Frankham 2008), and/or to enhance
populations in order to improve recreational activities such as fishing and hunting (Hansen et al.
2009; Burnside et al. 2016). However, captive-bred individuals can differ in their genetic and
phenotypic characteristics from wild ones, mainly because captivity and the selective pressures
associated with it strongly differ from natural environments (Frankham et al. 1986; Blanchet et al.
2008; Christie et al. 2016). Therefore, genetic admixture between wild populations and their
captive-bred conspecifics is a major conservation and wildlife management conundrum (Randi
2008; Araki and Schmid 2010).
Genetic admixture between captive-bred and wild individuals can strongly influence individual
fitness, notably through its consequences on behavioral, morphological and physiological traits
(Geiser and Ferguson 2001; Stoinski et al. 2003). Among these traits, dispersal (which is defined
here as all movements of individuals or propagules, potentially responsible for gene flow across
space and time; Ronce 2007) is a determinant mechanism for population dynamics (Clobert 2012).
Dispersal indeed plays a key role in the persistence of local populations and evolution of species'
spatial distribution, in particular because dispersal enables gene flow among populations, underlies
colonization and reduces extinction risk (Hanski 1998; Campbell Grant et al. 2010). Dispersal also
directly benefits individuals‘ fitness in response to environmental changes, kin competition, and
risk of inbreeding (Cressman and Křivan 2006). Moreover, differential dispersal (in terms of
probability to disperse, distance travelled and characteristics of destination sites) among
individuals varying in their genetic background (i.e. wild fish, admixed fish or descendants of
captive-bred fish) could condition the spatial distribution of admixture.
The individual determinants of dispersal have been widely studied (Clobert 2012; Lowe
and McPeek 2014) which permits generating predictions regarding how patterns of dispersal may
differ between captive-bred, wild and resulting admixed individuals. For instance, Dingemanse et
al. (2003) showed that individuals exhibiting higher exploratory behavior are more likely to
disperse, and do so over longer distances. Similarly, more aggressive and bold individuals should
disperse less, because they can successfully hold on to resources and to avoid costs of dispersal
141
(Cote et al. 2010; Hudina et al. 2015). Given that captive-bred individuals tend to show a lower
propensity for exploration (Robert et al. 1987; Johnsson and Abrahams 1991) and higher
aggression and dominance levels (Kelley et al. 2006; Frumkin et al. 2016), we may expect captive-
bred individuals to show a less pronounced dispersal behavior than wild ones. Alternatively, larger
or heavier individuals with higher growth rate may show higher dispersal propensity and move
over longer distances (Debeffe et al. 2014; Radinger and Wolter 2014; Dahirel et al. 2015).
Because captive-bred individuals generally exhibit higher growth rates and higher body mass
(Tymchuk and Devlin 2005), we may thus conversely expect captive-bred individuals to show a
more pronounced dispersal behavior than wild ones. To sum up, captive-bred and wild individuals
vary in many traits related to dispersal: clear predictions regarding differences between strains and
admixed individuals (i.e. those sharing the genome of both captive-bred and wild individuals), in
terms of dispersal propensity and distance, strongly depend on the relative importance of these
traits on dispersal behavior.
Studies comparing dispersal differences between wild and captive-bred strains focus
generally on immediate post-release movements. For instance, captive-bred fish released into a
natural system are less likely to disperse (Symons 1969; Jorgensen and Berg 1991), but those that
disperse do so over longer distances than wild ones (Bettinger and Bettoli 2002; Ebner and Thiem
2009). In birds, captive-bred individuals released in the wild tend to disperse less, and over shorter
distances than wild ones, probably because of both shorter life span and lower migration speed in
captive-bred individuals (Amar et al. 2008; Söderquist et al. 2013). To our knowledge however, no
study has yet focused on the direct link between genetic admixture and dispersal in a population in
which captive-bred strains have been implanted for a relatively long time, and in which admixture
is thus a part of the population dynamics.
Here, we used a headwater stream fish, the brown trout (Salmo trutta), to test whether or
not long-term admixture between wild individuals and captive-bred fish affects patterns of
dispersal. We focused on a population residing in a small mountain watershed that was heavily
stocked with captive-bred fish from 1972 to 1999, and in which we thoroughly sampled the whole
watershed. By combining this fine scale sampling design and a series of tools derived from
population genetics, we specifically aimed to (i) quantify genetic admixture of the supposedly wild
trout with captive-bred trout at the individual level, (ii) test how admixture was spatially
142
distributed within the river basin, and (iii) assess the effect of genetic admixture on individual
dispersal parameters (dispersal probability, dispersal direction and dispersal distances), while
simultaneously taking into account the effects of sex and body size on dispersal parameters (two
characteristics having been identified as major determinants of dispersal; Pusey 1987; Gutiérrez
and Menéndez 2003). We expected admixed fish and descendants of captive-bred fish to display
different dispersal behavior than wild fish, and dispersal traits divergence between groups to
increase with the increase in the proportion of allochtonous ancestry in the genome.
Material and Methods
Study area
The Neste d‘Oueil is a snow/rain-fed river from the French Pyrenees (Figure 1). Its source is at
1850m altitude, it confluences with the Neste d‘Oô River at 765m, after a 9.2 km course, and it is
situated ~500 km from the Atlantic Ocean mouth of the Garonne River (from which it is a
tributary). With a 1.8 % mean slope, its water flow varies from 0.4 m3.s
-1 to 1 m
3.s
-1. Its basin
drains 30.7 km2
and is composed of 14 tributaries, all shorter than 3 km in riparian distance. The
river basin is fragmented by ten obstacles, among which eight weirs (four of them are lower than
0.5 m and the other four are between 0.5 and ~1.5 m high), a culvert and a natural waterfall (1.5 m
high) (Figure 1). According to local managers and a recent telemetric survey (unpublished data),
all these obstacles are passable downstream, and upstream passage depends upon water flow
conditions. Highest obstacles are passable (upstream) on rare high flow conditions, whereas
smallest obstacles are passable during fall and spring normal high flows (Ovidio et al., 2015),
which –overall– suggests that gene flow can occur across the whole river catchment. The fish
community is dominated by sedentary brown trout (the population does not sustain anadromous
individuals as the river is situated far too upstream to allow for saltwater migrations), with some
rare bullhead individuals (Cottus gobio). The last brown trout stockings in the Neste d‘Oueil
occurred in 1999. Since 1972, these stockings were all done from a local trout hatchery (Soueich)
that is administrated by the regional angling association (Fédération Départementale pour la Pêche
et la Protection des Milieux Aquatiques de Haute Garonne) and used to stock most rivers in the
area. This hatchery was created in 1971, and fish from this hatchery originated from crosses
between Danish strains classically used in European hatcheries, and individuals from a
143
neighboring river basin (the Ger River). Unpublished studies showed that these hatchery fish form
a population being genetically distinct from wild populations from local rivers, even from the Ger
River itself. Until 1999, yearly stocking in the Neste d‘Oueil river basin mainly consisted in
releasing juvenile fish (young-of-the-year) and/or in placing incubated eggs across the entire river
basin, including the mainstream and tributaries. The local anglers associations in charge of the
stocking aimed at releasing fish evenly in the river basin, and did not target any particular
locations (personal communication). In some occasions (i.e. ~ once every two years), the
downstream part of the mainstream was additionally stocked with adults for recreational activity
(see Appendix S1).
Figure 4.1. (A) Maps representing the geographic situation of the Neste d’Oueil, as well as sampling sites,
and obstacles (natural: square dashed line, culverts: rectangle dashed line, artificial weirs lower than
0.5m: black rounded rectangles, artificial weirs higher than 0.5m: black rectangles) in the river basin. Pie
144
charts represent the proportions of individuals from each site that can be assigned to clusters “wild” (light
grey), “captive-bred” (black) and “mixed” (dark grey) as inferred from Q-values. The black arrow
indicates the location of the river mouth. (B) Classical bar plot for STRUCTURE results ("wild" cluster in
light grey and "captive-bred" cluster in black).
Field sampling
Brown trout were sampled in July 2014, using electric-fishing from 21 sites in total, 11 on the
mainstream (―NE‖ sites, Figure 1), and 10 on tributaries (―RU‖ sites, Figure 1). We failed to find
brown trout in four out of the 14 tributaries sampled despite intensive sampling efforts, which
indicated that they were probably fishless. We mainly sampled small individuals from the 1 year-
old class age. In total, we captured 715 individuals, with sample sizes ranging from 5 (RU3) to 72
(NE4), and with an average of 35 ± 19 individuals per site (Appendix S2 and S3). We additionally
sampled 48 fish from the Soueich trout hatchery to genetically characterize the captive-bred strain
used in this river, and hence to quantify genetic admixture with captive-bred trout. Each individual
was measured (total length in mm; see Appendix S2 for details), and a pelvic fin clip was taken for
genetic analyses. All individuals were released alive to their original sampling site.
Genotyping
Individual multilocus genotypes were obtained at a total of 17 markers. Among these 17 markers,
we used 16 microsatellite markers (BS31, One9, SsoSL311, SsoSL438, T3-13, Sfo1, Ssa064,
Ssa417, Ssa103, Ots515NWFSC, Ssa-60NVH, Ssa-TAP2a, Ssa-UBA, Ssa14, Ssa85DU and
SsoSL417; see Appendix S4 for details), and one sex-linked marker, Salmo-Sdy, which enables
determination of the sex of each individual (Quéméré et al. 2014). The 17 markers were assembled
in three PCR multiplexes.
Genomic DNA was extracted from the fin clips using a salt-extraction protocol (Aljanabi 1997).
The loci were amplified using the QIAGEN Multiplex PCR Kit (Qiagen, Valencia, CA, USA).
PCRs were carried out in a 10 µL final volume containing 5–20 ng of genomic DNA, 5 µL of 2X
QIAGEN Multiplex PCR Master Mix, and locus-specific optimized combination of primers under
the following conditions: 15 min at 95°C followed by 30 cycles of 1 min at 94°C, 1 min at 60°C
and 1 min at 72°C and finally followed by a 60 min elongation step at 72°C. Amplified fragments
145
were then separated on an ABI PRISM 3730 automated capillary sequencer. Allelic sizes were
called using GENEMAPPER v.4.0 (Applied Biosystems, Foster City, CA, USA).
Genetic diversity and differentiation
We performed preliminary analyses on all markers except for the one linked to sex. We first
investigated large allele drop-out and null alleles using Microchecker V2.2 (Van Oosterhout et al.
2004). We then tested for linkage disequilibrium between loci using FSTAT (Goudet 1995), and
we used LOSITAN (Antao et al. 2008) to determine whether or not some of these loci displayed
signs of natural selection, as suggested by previous studies (Blanchet et al. 2009; Keller et al.
2011). We did not detect significant linkage disequilibrium among loci but one of the 16 markers
(Ssa-UBA) displayed a strong deficit in heterozygosity, most likely because of the presence of null
alleles (Appendix S3). Two loci appeared to be under potential selection (Ots515 NWFSC and
Ssa14). Ssa-UBA, Ots515 NWFSC, and Ssa14 were therefore discarded from the database for
subsequent analyses. Overall, our final genetic dataset thus comprises 763 individuals genotyped
at 13 microsatellite markers (plus the sex marker Salmo-Sdy), with in total 1.48% of missing data.
Genetic diversity within each of the sampling sites was estimated over all loci by computing
unbiased expected heterozygosity (He) using GENETIX (Belkhir et al. 2004), standardized allelic
richness (i.e. the mean number of alleles corrected for the smaller sample size, Ar) after exclusion
of small sample locations (n<14; RU3, RU6, RU9 and RU12) using FSTAT, and the Fis index
using GENEPOP (Rousset 2008). Departure from Hardy-Weinberg equilibrium within each
sampling site and each locus was calculated using GENEPOP. Genetic differentiation was
assessed by computing global Fst over all sites and pairwise Fst between sites using the adegenet
R package (Jombart 2008).
Population structure and admixture extent
We assessed how S. trutta populations are genetically and spatially structured along the Neste
d‘Oueil, and whether there were differences in terms of spatial distribution between wild, captive-
bred descendants and admixed individuals. We performed genetic clustering using STRUCTURE
2.3.1 (Pritchard et al. 2000) with the admixture model and the correlated allele frequency model,
without prior population information. Runs were performed with a burn-in period of 200.000 and
146
200.000 subsequent MCMC repetitions. The number K of clusters ranged from 1 to 10 (5 runs
each). Log-likelihood plots and ΔK statistics were obtained via the Evanno‘s method (Evanno et
al. 2005) using STRUCTURE HARVESTER (Earl and vonHoldt 2012), and they were used to
infer the optimal K-value. Twenty runs were then performed with this optimal K-value and the ten
best runs (the ones with highest LnP(D) values) were compiled using CLUMPP (Jakobsson and
Rosenberg 2007) to get final averaged individual Q-values. Individuals were assigned to the
cluster with the greatest Q-value, provided that value exceeded 0.7, as this has been done for other
salmonid species (Hansen et al. 2001; Vähä and Primmer 2005; Valiquette et al. 2014). Individuals
with intermediate Q-values (i.e between 0.3 and 0.7) were considered genetically admixed.
Graphical displays of STRUCTURE plots were generated using DISTRUCT software (Rosenberg
2003). We directly used the probability of assignment (individual Q-value) to the cluster
containing all individuals from the Soueich trout hatchery as an estimate of individual genetic
admixture with the captive-bred strain (Hansen et al. 2001).
In order to further describe the spatial structure of strains in the river basin, we tested the
relationship between site distance to the river mouth (confluence between the Neste D‘Oô and the
One Rivers; Figure 1) and admixture level (averaged at the population level and log-transformed).
We also tested whether or not admixture level and distance to the mouth (and the resulting
interaction term) were significantly related to mean allelic richness and population differentiation
(mean Fst) respectively. Relationships were tested using linear models, and all variables were
standardized. Finally, we tested to what extent geographic isolation and fragmentation by weirs
affected the spatial distribution of strains by testing the relationships between difference in mean
admixture between pairs of sites, pairwise riparian distance and number of obstacles between pairs
of sites, using multiple regressions on distance matrices (MRDM; Smouse et al. 1986) coupled
with regression commonality analyses (CA; Prunier et al. 2015).
Inferring dispersal from first generation migrants
We then aimed at identifying individual dispersal events by inferring ―first generation migrants‖
(i.e. F0 immigrants) using GENECLASS 2 (Piry et al. 2004). We used Paetkau‘s method (Paetkau
et al. 1995) to assign ―first generation migrants‖ to their population of origin which involves
calculating the expected frequency of each individual's genotype in each population (product of
147
expected genotype frequency at each locus, based on the observed distributions of alleles) and
subsequent assignment of each individual to the population where its expected genotype frequency
is highest. We tested the null hypothesis that an individual was born in the population in which it
was sampled using a Monte Carlo resampling method (Rannala and Mountain 1997; Cornuet et al.
1999). Rejection of the null hypothesis indicated an individual having dispersed from one site to
another. The probability threshold for inferring first generation migrants was set to 0.01.
We tested the effect of level of admixture, sex and body length of each individual on four dispersal
parameters (response variables): (i) the individual dispersal probability (obtained by
transformation of the binomial variable ―dispersal vs. non-dispersal‖ using the predict() R
function), (ii) the individual dispersal distance (i.e. geographic river distance between the site of
origin and the site of destination), (iii) the individual dispersal direction (i.e. upstream or
downstream-directed regarding the water flow), and (iv) the probability of stream type of the
destination site (i.e. transformed binomial variable ―dispersal to a site of the mainstream vs. to a
tributary‖ using the R predict() function). We used Generalized Linear Models (GLM; Gaussian or
Binomial error terms depending on the response variable) and we standardized all continuous
variables. We integrated the quadratic term for the level of admixture to test for potential non-
linear relationships. For each of the four dependent variables, we then used a model selection
procedure based on the Akaike Information Criteria (AIC) to identify the most parsimonious set of
predictors.
Results
Genetic diversity and differentiation
Expected heterozygosity over all loci (He) ranged from 0.42 to 0.74 (mean = 0.65 ± 0.09),
standardized mean allelic richness (Ar) varied from 4.49 to 9.39 (6.24 ± 1.21), and Fis within each
site ranged from -0.06 to 0.15 (0.053 ± 0.047) (Appendix S4). Eight out of 21 populations
significantly deviated from Hardy-Weinberg equilibrium (Appendix S4). The mean genetic
differentiation (Fst) estimated across sites and loci was 0.12 (± 0.07), and pairwise Fst values
between sites ranged from 0 between NE8 and NE10 (geographically close sites) to 0.21 between
RU1 and RU16 (geographically distant sites) with an average of 0.070 (± 0.023). Finally, the sex
148
ratio in each site did not significantly differ from an equal ratio within sites (i.e. 50:50; 2=0.55,
d.f. = 1, p = 0.46).
Population structure and admixture extent
Individuals were assigned to two main genetic clusters. The first cluster (Cluster A, in light gray in
Figure 1) mainly regrouped individuals from the mainstream and from some of the tributaries
(RU4, RU7, RU9, RU10, RU11 and RU12). All individuals from RU1, RU3, RU6, RU16 and
from the Soueich trout farm were assigned to the second cluster (Cluster B, in black in Figure 1),
as well as a few individuals from NE1, RU4, RU7 and RU10. We therefore considered the first
cluster as the ―natural‖ population of the basin (hereafter ―wild cluster‖), and the second cluster as
derived from past stocking activities (hereafter ―captive-bred cluster‖). We found that 79 % of the
fish were most likely assigned to the ―wild cluster‖ (566 individuals with Q-value < 0.3), 11 %
were most likely assigned to the ―captive-bred cluster‖ (77individuals with Q-value > 0.7), while
10% were equally assigned to both the wild and captive-bred clusters. Moreover, within the
captive-bred cluster, subclustering showed two different clusters (Appendix S5), with all
individuals from the trout farm on the one hand and individuals caught within the river basin on
the other hand, hence suggesting genetic drift and/or confirming a partial admixture with the wild
population. We found a positive correlation between distance to the river mouth and level of
genetic admixture with captive-bred trout (r = 0.55, d.f. = 19, P-value < 0.01), indicating that
individuals strongly assigned to the captive-bred cluster were mostly found on a few upstream
tributaries (Figure 1).
Patterns of genetic diversity and admixture
Regarding patterns of allelic richness, we found a significant interaction term between admixture
level (at the site level) and distance to the river mouth (Appendix S6, allelic richness). This
indicated that, for an intermediate level of admixture, allelic richness was higher in upstream than
in downstream sites (Appendix S6; see also Appendix S7 for the interpretation of model
parameters in presence of a first-order interaction). This spatial trend held true at a low level of
admixture (wild individuals) but not at a high level of admixture (Figure 2): downstream sites
149
associated with high admixture rates (notably RU16; Figures 1 and 2) were hence responsible for
an inversion in the ―natural‖ downstream increase in genetic diversity.
Figure 4.2. Relationship between mean allelic richness and distance to the river mouth, depending on mean
level of admixture at each site. In light grey, line for low admixture; in grey, line for average admixture; in
black, line for high admixture.
We further found a significant relationship between mean Fst measured at the site level and level
of admixture: populations with a high proportion of fish assigned to the captive-bred cluster
exhibited on average higher genetic differentiation (Appendix S6, Fst). Finally, we found that the
number of obstacles significantly explained differences in admixture between sites (beta = 0.472,
p-value = 0.004; Unique contribution = 0.109, Common contribution = 0.036), whereas riparian
distance alone did not (beta = -0.127, p-value = 0.314; Unique contribution = 0.008, Common
contribution = 0.036).
Inferring dispersal from first generation migrants
GENECLASS inferred 43 ―first generation migrants‖ (24 females and 19 males) with a probability
higher than 0.99, meaning that 5.6 % of individuals dispersed. Twenty individuals dispersed
upstream, 23 dispersed downstream. Thirty-one of these migrants dispersed towards a site of the
mainstream whereas 12 dispersed towards a tributary. Individual dispersal distances ranged from
322.8 to 8062.1 m with a median of 1921.3 m (± 2124.4).
150
The best model retained to explain individual dispersal probability included both genetic
admixture with captive-bred trout (simple term only) and trout body length (Table 1, A). As shown
in Appendix S8_A, dispersal probability was positively correlated with the probability of
assignment to the captive-bred cluster (individuals with a high proportion of allochtonous ancestry
were more likely to undergo a dispersal event). We additionally found a slight tendency for
smaller individuals to be more likely to disperse. The model retained to explain individual
dispersal distance also comprised body length as well as the simple and quadratic terms of genetic
admixture (Table 1, B). Individuals strongly assigned either to the wild or to the captive-bred
cluster tended to disperse over longer distances than individuals with mixed assignments
(Appendix S8_B). Moreover, smaller individuals tended to disperse over longer distances. The
best model explaining destination stream type comprised genetic admixture only (Table 1, D). The
probability to disperse towards a site of the mainstream rather than towards a tributary was
negatively correlated with the probability of being assigned to the captive-bred cluster. Finally, the
null (intercept only) model was the best at explaining individual dispersal direction (Table 1, C),
indicating that factors other than sex, body length and genetic admixture could explain individual
variations in dispersal direction. Interestingly, sex was never retained in final simplified models,
indicating no differences between males and females in dispersal traits.
A B C D
Dispersal probability Dispersal Distance Dispersal direction Destination stream level
β SE P β SE P β SE P β SE P
Intercept -2.800 0.166 < 0.001** < 10e-6 0.135 1.000 0.191 0.310 0.538 1.062 0.392 0.007**
Sex / / / /
Body length -0.131 0.152 0.389 -0.338 0.151 0.031* / /
Admixture 0.288 0.131 0.028* -0.957 0.640 0.143 / -0.966 0.365 0.008**
Admixture 2 / 1.274 0.628 0.049* / /
Table 4.1. Results from the final models retained for testing the role of sex, body length and admixture
(along with the intercept) on the individual dispersal probability (A), the individual dispersal distance (B),
the individual dispersal direction (C) and the type (i.e. mainstream or tributary) of the destination site of
first generation migrants (D). Variables that were not retained in the final model are indicated by a slash
bar. *: p-value < 0.05; **: p-value < 0.01.
151
Discussion
Patterns of genetic admixture and genetic diversity at the basin scale
Although stocking was stopped more than 16 years ago (i.e. 6-8 trout generations ago), we found
that 21 % of individuals caught in the watershed were at least partially assigned to the captive-bred
cluster. In particular, half of these individuals had a high assignment probability to the captive-
bred cluster (> 0.7), suggesting either pure descendance from the captive-bred strain or back-
crosses between admixed individuals and descendants from captive-bred fish. These results were
difficult to compare to other systems because admixture rates between captive-bred and wild trout
populations (and hence outcomes after stocking is stopped) are highly variable (Cagigas et al.
1999; Perrier et al. 2013).
Interestingly, the distribution of captive-bred genotypes in the watershed was not homogeneous.
More precisely, some small tributaries, notably those situated in upstream parts of the basin, were
exclusively populated with fish strongly assigned to the captive-bred cluster, whereas in the main
river, fish strongly assigned to the captive-bred cluster were rare despite the homogeneous
stocking effort in the watershed. This suggests that past stocking events significantly influenced
the spatial distribution of alleles in this river basin. The persistence of fish strongly assigned to the
captive-bred cluster in some tributaries could be due to the fact that these stretches were fishless or
at very low density before stocking occurred: competition with the native strain was probably low,
thus facilitating their settlement and reproductive success, enabling them to co-occur in parapatry
with the original wild population. We also showed that obstacles partly explained differences in
admixture between sites, and thus also probably contribute to favoring spatial segregation of the
two strains in the basin. Indeed, some of these obstacles are difficult to cross upstreamwards at
normal flow conditions, which may limit hybridization between wild and captive-bred individuals
upstream of these obstacles. Nevertheless, additional information as to the historical spatial
distribution of trout in this river system would be required to shed light on this spatial pattern.
Our results also suggested that past stocking activities and admixture with captive-bred trout
affected spatial patterns of genetic diversity and genetic differentiation. We notably found an
overall downstream decrease in allelic richness when levels of admixture were null to moderate
152
(Fig. 2 and Appendix S6), contrary to the traditional expectation of a downstream increase in
genetic diversity (Morrissey and de Kerckhove 2009; Paz-Vinas et al. 2015). This pattern has
already been observed though (Cyr and Angers 2011; Conti et al. 2015) and, it could be explained
by upstream-biased gene flow, higher effective population sizes in upstream stretches, and/or a
historical colonization of the river that began upstream and ended downstream (Paz-Vinas et al.
2015). Conversely, allelic richness slightly increased downstream for high levels of admixture,
indicating that the distribution of allochtonous genotypes introduced with stocking events differed
from the distribution of wild ones. Overall, our findings demonstrated that stocking, even when it
occurred several generations ago, can strongly affect spatial patterns of allelic richness while
increasing genetic differentiation of populations carrying a high proportion of allochtonous
genotypes (Marie et al. 2010; Valiquette et al. 2014), highlighting the necessity for stocking events
to be taken into account in riverscape genetics studies (Prunier et al. 2018).
Captive breeding and genetic admixture affect patterns of dispersal
Although we expected a male-biased dispersal or at least higher male mobility (McGinnity et al.
2003) due to the polygamous and/or polyandrous mating system of trout and to the strong
competition for mates, we here found no evidence for an effect of sex on dispersal. Possible
explanations could be a lack of statistical power, the lack of sex-biased dispersal for immature
(1+)-trout as mainly sampled, or the actual absence of sex bias in dispersal at such a small spatial
scale. We did not find any significant correlation between fish size and dispersal probability either,
although younger, and therefore smaller individuals, are generally more likely to disperse because
their territory is not yet established (Andreu and Barba 2006; Gachot-Neveu et al. 2009).
Nevertheless, the ―first generation migrant‖ assignment approach indicated that smaller fish
disperse further than larger ones, a pattern already observed in riverine fish (Skalski and Gilliam
2000). Since shorter dispersal can enhance survival by reducing mortality risk (Johnsson et al.
1999), larger and more dominant fish may chose not to disperse over long distances, forcing the
smaller ones to disperse further (Vøllestad et al. 2012). It is noteworthy that previous studies on
brown trout populations regularly mentioned environmental variables (Cucherousset et al. 2005),
fish densities (Olsson et al. 2006) or physiological status (Rustadbakken et al. 2004) as major
determinants of dispersal, which were not considered in our study.
153
On the contrary, a constant driver of dispersal probability and dispersal distance in our study was
the probability of assignment to the captive-bred cluster. Indeed, we found that: (i) dispersal
probability was affected by the probability of assignment to the captive-bred cluster, and (ii)
dispersal distances covered by dispersing individuals were lower in individuals with a mixed
assignment than in individuals assigned either to the wild or captive-bred cluster. It is noteworthy
that this later relationship was not due to the fact that captive-bred individuals were confined to
tributaries (that we might expect to be farther away from a neighboring population than a
population from the mainstream) since tributary-neighboring populations were actually not
significantly further apart than mainstream-neighboring populations (mean distance to neighboring
site in the mainstream: 733m vs. mean distance to neighboring site in tributaries: 826m; t-test, t = -
0.63, df = 17.25, p = 0.53). Our results further revealed that individuals with genotypes closer to
those of the trout farm were more likely to disperse towards a tributary whereas individuals
strongly assigned to the wild cluster were more likely to disperse towards a site of the mainstream,
which may contribute to maintaining the spatial segregation of strains observed in the watershed.
Importantly, these findings were confirmed using another independent analysis based on dispersal
measured by reconstructing full sibling families (see Appendix S9 for further details). To our
knowledge, our study is one of the first to reveal such a pattern of ―admixture-biased dispersal‖,
which has implications for understanding population dynamics in an environment being -or having
been- subject to stocking by fish of non-native origin.
This ―admixture-biased dispersal‖ can have diverse explanations. For instance, admixed
individuals may show differences in behavioral dominance and/or body size (Edelaar and Bolnick
2012). However, we did not identify any link between admixture and size, a main trait of
dominance (Miller and Frey 1972). Alternatively –and non-exclusively–, the observed spatial
distribution of admixture levels could result from differences in fitness between fish with a mixed
assignment to the wild and captive-bred clusters (admixed fish per se) and pure (wild or captive-
bred descendants) individuals (Johnson et al. 2010). Indeed, the average hybrid phenotypes have
in many cases been shown to lie outside the phenotypic range of the parental ones, providing
hybrid vigor or conversely hybrid depression (Facon et al. 2005; Rasmussen et al. 2012). These
phenomena have been documented in salmonids (Wollebæk et al. 2012), although they are still
ambiguous (McClelland and Naish 2007). Thus, predictions on the long-term genetic
154
consequences of stocking is still a challenge, and the link between our results and these differential
fitness outcomes remains unclear (Harrison et al. 2005), thus adding a hypothesis for explaining
our results. However, this pattern has many counterexamples in which admixed individuals show
strong fitness increases enabling them to be better colonizers (Drake 2006; Keller and Taylor
2010).
Local selection pressures (Edelaar and Bolnick 2012) and/or differential habitat matching choice
between the two strains (Edelaar et al. 2008) could also help explain both dispersal differences
between strains and the observed heterogeneous spatial distribution of strains in the watershed. In
habitat matching choice, individuals with a given phenotype aim to settle in the environment in
which their fitness is optimized (Edelaar et al. 2008). Conversely to what is usually expected as a
consequence of high gene flow (i.e. lower genetic differentiation among demes), matching habitat
choice through dispersal could increase and/or maintain among-habitat divergence between the
two different strains (Bolnick and Nosil 2007). In that case, this may contribute to maintain –
together with fragmentation by weirs or natural obstacles– the spatial segregation between the
―captive-bred‖ and the ―wild‖ populations. Evaluating the relative relevance of such a hypothesis
(compared to fragmentation) would require further investigations on the habitat characteristics
(tributaries vs. mainstream) used by captive-bred and wild individuals and on the relative fitness of
―captive-bred‖ and ―wild‖ individuals in their respective habitats.
Conclusions
We demonstrated that parameters of dispersal can be affected by the proportion of allochtonous
alleles brought by stocking activity, and that these effects of admixture can be observed long after
stocking activities ceased. Although the mechanisms sustaining differences in dispersal along this
gradient of admixture are still unknown, this finding has important implications for understanding
and predicting the spread, distribution and maintenance of allochtonous alleles in wild
populations. For instance, we here revealed a strong spatial segregation between ―wild‖ and
―captive-bred‖ strains at the watershed level. This heterogeneous spatial distribution could be
maintained over time because of the differences in dispersal direction observed between wild and
descendants of captive-bred fish, which in the long term could limit the spread of allochtonous
alleles. However, this distribution could also be the result of competitive exclusion, where the
155
upstream populations act as sources of allochtonous alleles, which have trouble colonizing the
downstream sites in which the wild strain is already present. We call for future studies completing
these important findings and testing for underlying mechanisms. Indeed, it would be of interest to
test whether differential dispersal parameters can affect the spatial dynamics of admixture in a
watershed.
Data accessibility
Raw data are deposited on figshare. DOI: 10.6084/m9.figshare.6886670.
Authors' contributions
SB, NP and LT designed the experiment and coordinated the study; KSP, SB, NP, LT, OP, GL and
CV conducted sampling; KSP, GL and CV carried out the experimental lab work; KSP and JGP
ran the statistical analyses; KSP, JGP, SB, NP, LT, OP, GL and CV interpreted the data. KSP, SB,
and JGP wrote the first draft of the manuscript. NP, LT, OP, GL and CV read, commented and
corrected the initial draft, and all authors gave final approval for publication.
Acknowledgments
We warmly thank the AAPPMA of Luchon (Jean Lérimé, Daniel Estrade and colleagues) and
ECOGEA (Jean-Marc Lascaux, Philippe Baran and colleagues) for field sampling support. The
genetic data were generated at the molecular genetics technical facilities of the Genopole Midi-
Pyrénées (Toulouse, France). This work was undertaken at SETE, which forms part of the
―Laboratoire d‘Excellence‖ (LABEX) entitled TULIP (ANR-10-LABX-41).
Funding
This work was funded by the French Agency for Biodiversity (n° 136406) and Electricité De
France (n°8619-5920011137).
Conflict of Interest Statement
We have no competing interests and we declare no conflict of interests.
156
References
Aljanabi S (1997) Universal and rapid salt-extraction of high quality genomic DNA for PCR-
based techniques. Nucleic Acids Res 25:4692–4693.
Amar A, Arroyo BE, Bretagnolle V (2008) Post-fledging dependence and dispersal in hacked and
wild Montagu‘s Harriers Circus pygargus. Ibis 142:21–28.
Andreu J, Barba E (2006) Breeding dispersal of Great Tits Parus major in a homogeneous habitat:
effects of sex, age, and mating status. Ardea-Wageningen 94:1– 45.
Antao T, Lopes A, Lopes RJ, et al (2008) LOSITAN: A workbench to detect molecular adaptation
based on a Fst-outlier method. BMC Bioinformatics 9:323.
Araki H, Schmid C (2010) Is hatchery stocking a help or harm? Aquaculture 308:S2–S11.
Bettinger JM, Bettoli PW (2002) Fate, dispersal, and persistence of recently stocked and resident
Rainbow Trout in a Tennessee tailwater. North Am J Fish Manag 22:425–432.
Blanchet S, Páez DJ, Bernatchez L, Dodson JJ (2008) An integrated comparison of captive-bred
and wild Atlantic salmon (Salmo salar): Implications for supportive breeding programs.
Biol Conserv 141:1989–1999.
Bolnick DI, Nosil P (2007) Natural selection in populations subject to a migration load. Evolution
61:2229–2243.
Burnside RJ, Collar NJ, Scotland KM, Dolman PM (2016) Survival rates of captive-bred Asian
Houbara Chlamydotis macqueenii in a hunted migratory population. Ibis 158:353–361.
Cagigas ME, Vazquez E, Blanco G, Sanchez JA (1999) Genetic effects of introduced hatchery
stocks on indigenous brown trout (Salmo trutta L.) populations in Spain. Ecol Freshw Fish
8:141–150.
Campbell Grant EH, Nichols JD, Lowe WH, Fagan WF (2010) Use of multiple dispersal pathways
facilitates amphibian persistence in stream networks. Proc Natl Acad Sci 107:6936–6940.
Christie MR, Marine ML, Fox SE, et al (2016) A single generation of domestication heritably
alters the expression of hundreds of genes. Nat Commun 7:10676.
Clobert J (ed) (2012) Dispersal ecology and evolution, 1st ed. Oxford University Press, Oxford
Conti L, Comte L, Hugueny B, Grenouillet G (2015) Drivers of freshwater fish colonisations and
extirpations under climate change. Ecography 38:510–519.
Cornuet JM, Piry S, Luikart G, et al (1999) New methods employing multilocus genotypes to
select or exclude populations as origins of individuals. Genetics 153:1989–2000
Cote J, Clobert J, Brodin T, et al (2010) Personality-dependent dispersal: characterization,
ontogeny and consequences for spatially structured populations. Philos Trans R Soc B Biol
Sci 365:4065–4076.
Cressman R, Křivan V (2006) Migration dynamics for the Ideal Free Distribution. Am Nat
168:384–397.
Cucherousset J, Ombredane D, Charles K, et al (2005) A continuum of life history tactics in a
brown trout (Salmo trutta) population. Can J Fish Aquat Sci 62:1600–1610.
Cyr F, Angers B (2011) Historical process lead to false genetic signal of current connectivity
among populations. Genetica 139:1417–1428.
Dahirel M, Olivier E, Guiller A, et al (2015) Movement propensity and ability correlate with
ecological specialization in European land snails: comparative analysis of a dispersal
syndrome. J Anim Ecol 84:228–238.
157
Debeffe L, Morellet N, Bonnot N, et al (2014) The link between behavioural type and natal
dispersal propensity reveals a dispersal syndrome in a large herbivore. Proc R Soc B Biol
Sci 281:20140873–20140873.
Dingemanse NJ, Both C, van Noordwijk AJ, et al (2003) Natal dispersal and personalities in great
tits (Parus major). Proc R Soc B Biol Sci 270:741–747.
Drake JM (2006) Heterosis, the catapult effect and establishment success of a colonizing bird. Biol
Lett 2:304–307.
Earl DA, vonHoldt BM (2012) STRUCTURE HARVESTER: a website and program for
visualizing STRUCTURE output and implementing the Evanno method. Conserv Genet
Resour 4:359–361.
Ebner BC, Thiem JD (2009) Monitoring by telemetry reveals differences in movement and
survival following hatchery or wild rearing of an endangered fish. Mar Freshw Res 60:45.
Edelaar P, Bolnick DI (2012) Non-random gene flow: an underappreciated force in evolution and
ecology. Trends Ecol Evol 27:659–665.
Edelaar P, Siepielski AM, Clobert J (2008) Matching habitat choice causes directed gene flow: a
neglected dimension in evolution and ecology. Evolution 62:2462–2472.
Evanno G, Regnaut S, Goudet J (2005) Detecting the number of clusters of individuals using the
software structure: a simulation study. Mol Ecol 14:2611–2620.
Geiser F, Ferguson C (2001) Intraspecific differences in behaviour and physiology: effects of
captive breeding on patterns of torpor in feathertail gliders. J Comp Physiol [B] 171:569–
576.
Facon B, Jarne P, Pointier JP, David P (2005) Hybridization and invasiveness in the freshwater
snail Melanoides tuberculata: hybrid vigour is more important than increase in genetic
variance: Hybrid vigour in an invasive snail. J Evol Biol 18:524–535.
Frankham R (2008) Genetic adaptation to captivity in species conservation programs. Mol Ecol
17:325–333.
Frankham R, Hemmer H, Ryder OA, et al (1986) Selection in captive populations. Zoo Biol
5:127–138.
Frumkin NB, Wey TW, Exnicios M, et al (2016) Inter-annual patterns of aggression and pair
bonding in captive American flamingos (Phoenicopterus ruber): Flamingo Behavior. Zoo
Biol 35:111–119.
Gachot-Neveu H, Lefevre P, Roeder J-J, et al (2009) Genetic detection of sex-biased and age-
biased dispersal in a population of wild carnivore, the Red Fox, Vulpes vulpes. Zoolog Sci
26:145–152.
Goudet J (1995) FSTAT (Version 1.2): A computer program to calculate F-statistics. J Hered
86:485–486.
Gutiérrez D, Menéndez R (2003) Patterns in the distribution, abundance and body size of carabid
beetles (Coleoptera: Caraboidea) in relation to dispersal ability. J Biogeogr 24:903–914.
Hansen MM, Fraser DJ, Meier K, Mensberg K-LD (2009) Sixty years of anthropogenic pressure: a
spatio-temporal genetic analysis of brown trout populations subject to stocking and
population declines. Mol Ecol 18:2549–2562.
Hansen MM, Nielsen EE, Bekkevold D, Mensberg K-LD (2001) Admixture analysis and stocking
impact assessment in brown trout (Salmo trutta), estimated with incomplete baseline data.
Can J Fish Aquat Sci 58:1853–1860.
Hanski I (1998) Metapopulation dynamics. Nature 396:41–49.
158
Harrison JF, Taylor Jr. OR, Hall HG (2005) The flight physiology of reproductives of Africanized,
European, and hybrid Honeybees (Apis mellifera). Physiol Biochem Zool 78:153–162.
Hudina S, Žganec K, Hock K (2015) Differences in aggressive behaviour along the expanding
range of an invasive crayfish: an important component of invasion dynamics. Biol
Invasions 17:3101–3112.
Jakobsson M, Rosenberg NA (2007) CLUMPP: a cluster matching and permutation program for
dealing with label switching and multimodality in analysis of population structure.
Bioinformatics 23:1801–1806.
Johnson JR, Fitzpatrick BM, Shaffer HB (2010) Retention of low-fitness genotypes over six
decades of admixture between native and introduced tiger salamanders. BMC Evol Biol
10:147.
Johnsson JI, Abrahams MV (1991) Interbreeding with domestic strain increases foraging under
threat of predation in juvenile Steelhead Trout (Oncorhynchus mykiss): An Experimental
Study. Can J Fish Aquat Sci 48:243–247.
Johnsson JI, Nobbelin F, Bohlin T (1999) Territorial competition among wild brown trout fry:
effects of ownership and body size. J Fish Biol 54:469–472.
Jombart T (2008) adegenet: a R package for the multivariate analysis of genetic markers.
Bioinformatics 24:1403–1405.
Jorgensen J, Berg S (1991) Stocking experiments with 0 + and 1 + trout parr, Salmo trutta L., of
wild and hatchery origin: 2. Post-stocking movements. J Fish Biol 39:171–180.
Keller SR, Taylor DR (2010) Genomic admixture increases fitness during a biological invasion:
Admixture increases fitness during invasion. J Evol Biol 23:1720–1731.
Kelley JL, Magurran AE, Macías García C (2006) Captive breeding promotes aggression in an
endangered Mexican fish. Biol Conserv 133:169–177.
Lowe WH, McPeek MA (2014) Is dispersal neutral? Trends Ecol Evol 29:444–450.
Marie AD, Bernatchez L, Garant D (2010) Loss of genetic integrity correlates with stocking
intensity in brook charr (Salvelinus fontinalis): Impacts of stocking on salvelinus fontinalis.
Mol Ecol 19:2025–2037.
McClelland EK, Naish KA (2007) What is the fitness outcome of crossing unrelated fish
populations? A meta-analysis and an evaluation of future research directions. Conserv
Genet 8:397–416.
McGinnity P, Prodohl P, Ferguson A, et al (2003) Fitness reduction and potential extinction of
wild populations of Atlantic salmon, Salmo salar, as a result of interactions with escaped
farm salmon. Proc R Soc B Biol Sci 270:2443–2450.
Miller RJ, Frey DF (1972) The establishment of dominance relationships in the blue gourami,
Trichogaster Trichopter Us (Pallas). Behaviour 42:8–60.
Morrissey MB, de Kerckhove DT (2009) The maintenance of genetic variation due to asymmetric
gene flow in dendritic metapopulations. Am Nat 174:875–889.
Olsson IC, Greenberg LA, Bergman E, Wysujack K (2006) Environmentally induced migration:
the importance of food. Ecol Lett 9:645–651.
Paetkau D, Calvert W, Stirling I, Strobeck C (1995) Microsatellite analysis of population structure
in Canadian polar bears. Mol Ecol 4:347–354.
Paz-Vinas I, Loot G, Stevens VM, Blanchet S (2015) Evolutionary processes driving spatial
patterns of intraspecific genetic diversity in river ecosystems. Mol Ecol 24:4586–4604.
159
Perrier C, Guyomard R, Bagliniere J-L, et al (2013) Changes in the genetic structure of Atlantic
salmon populations over four decades reveal substantial impacts of stocking and potential
resiliency. Ecol Evol 3:2334–2349.
Piry S, Alapetite A, Cornuet J-M, et al (2004) GENECLASS2: a software for genetic assignment
and first-generation migrant detection. J Hered 95:536–539.
Pritchard JK, Stephens M, Donnelly P (2000) Inference of population structure using multilocus
genotype data. Genetics 155:945–959
Prunier JG, Colyn M, Legendre X, et al (2015) Multicollinearity in spatial genetics: separating the
wheat from the chaff using commonality analyses. Mol Ecol 24:263–283.
Prunier JG, Dubut V, Loot G, et al (2018) The relative contribution of river network structure and
anthropogenic stressors to spatial patterns of genetic diversity in two freshwater fishes: A
multiple-stressors approach. Freshw Biol 63:6–21.
Pusey AE (1987) Sex-biased dispersal and inbreeding avoidance in birds and mammals. Trends
Ecol Evol 2:295–299.
Quéméré E, Perrier C, Besnard A-L, et al (2014) An improved PCR-based method for faster sex
determination in brown trout (Salmo trutta) and Atlantic salmon (Salmo salar). Conserv
Genet Resour 6:825–827.
Radinger J, Wolter C (2014) Patterns and predictors of fish dispersal in rivers. Fish Fish 15:456–
473.
Randi E (2008) Detecting hybridization between wild species and their domesticated relatives.
Mol Ecol 17:285–293.
Rannala B, Mountain JL (1997) Detecting immigration by using multilocus genotypes. Proc Natl
Acad Sci 94:9197–9201.
Rasmussen JB, Robinson MD, Hontela A, Heath DD (2012) Metabolic traits of westslope
cutthroat trout, introduced rainbow trout and their hybrids in an ecotonal hybrid zone along
an elevation gradient: metabolic traits of trout hybrids. Biol J Linn Soc 105:56–72.
Robert S, Dancosse J, Dallaire A (1987) Some observations on the role of environment and
genetics in behaviour of wild and domestic forms of Sus scrofa (European wild boars and
domestic pigs). Appl Anim Behav Sci 17:253–262.
Ronce O (2007) How does it feel to be like a rolling stone? Ten questions about dispersal
evolution. Annu Rev Ecol Evol Syst 38:231–253.
Rosenberg NA (2003) DISTRUCT: a program for the graphical display of population structure:
program note. Mol Ecol Notes 4:137–138.
Rousset F (2008) genepop‘007: a complete re-implementation of the genepop software for
Windows and Linux. Mol Ecol Resour 8:103–106.
Rustadbakken A, L‘Abee-Lund JH, Arnekleiv JV, Kraabol M (2004) Reproductive migration of
brown trout in a small Norwegian river studied by telemetry. J Fish Biol 64:2–15.
Seddon PJ, Armstrong DP, Maloney RF (2007) Developing the science of reintroduction biology.
Conserv Biol 21:303–312.
Skalski GT, Gilliam JF (2000) Modeling diffusive spread in a heterogeneous population: a
movement study with stream fish. Ecology 81:1685–1700.
Smouse PE, Long JC, Sokal RR (1986) Multiple regression and correlation extensions of the
mantel test of matrix correspondence. Syst Zool 35:627.
Söderquist P, Gunnarsson G, Elmberg J (2013) Longevity and migration distance differ between
wild and hand-reared mallards Anas platyrhynchos in Northern Europe. Eur J Wildl Res
59:159–166.
160
Stoinski TS, Maple TL, Beck BB, Bloomsmith MA (2003) A behavioral comparison of captive-
born, reintroduced golden lion tamarins and their wild-born offspring. Behaviour 140:137–
160.
Symons PEK (1969) Greater dispersal of wild compared with hatchery-reared juvenile atlantic
salmon released in streams. J Fish Res Board Can 26:1867–1876.
Tymchuk WE, Devlin RH (2005) Growth differences among first and second generation hybrids
of domesticated and wild rainbow trout (I). Aquaculture 245:295–300.
Vähä J-P, Primmer CR (2005) Efficiency of model-based Bayesian methods for detecting hybrid
individuals under different hybridization scenarios and with different numbers of loci:
genetic detection of hybrids. Mol Ecol 15:63–72.
Valiquette E, Perrier C, Thibault I, Bernatchez L (2014) Loss of genetic integrity in wild lake trout
populations following stocking: insights from an exhaustive study of 72 lakes from
Québec, Canada. Evol Appl 7:625–644.
Van Oosterhout C, Hutchinson WF, Wills DPM, Shipley P (2004) micro-checker: software for
identifying and correcting genotyping errors in microsatellite data. Mol Ecol Notes 4:535–
538.
Vøllestad LA, Serbezov D, Bass A, et al (2012) Small-scale dispersal and population structure in
stream-living brown trout (Salmo trutta) inferred by mark–recapture, pedigree
reconstruction, and population genetics. Can J Fish Aquat Sci 69:1513–1524.
Wollebæk J, Røed KH, Brabrand Å, Heggenes J (2012) Interbreeding of genetically distinct native
brown trout (Salmo trutta) populations designates offspring fitness. Aquaculture 356–
357:158–168.
161
Supplementary material for Chapter IV.
Appendix S4.1: Pieces of information about stocking events in the Neste d’Oueil River gathered
from local angling associations, in addition to informal interviews.
Year Vibert Boxes (several
thousand eggs each)
Larvae with
yolk-sac
Juveniles 0+
(7/10 cm) Adults (size in cm)
1987 - - 4000 350 (24/30)
1988 - - 10000 400 (24/30 and 40/50)
1989 40 20000 - 1200 (22/25)
1990 24 - - -
1991 20 8000 - 400 (20/25)
1992 46 - - -
1994 15 - - -
1995 - - 50000 -
1996 16 - - -
1997 - - - -
1998 20 - - 300 (22/25)
1999 - - 8000 -
162
Appendix S4.2: Histogram of individual sizes (A, in mm), and distribution of individual sizes
among sites (B) and between the mainstream and tributaries (C).
Average size varies among sites (Df = 20, F-value = 2.343, p-value = 0.000831). However, sizes
do not differ significantly between the mainstream and the tributaries, (Df = 1, F-value = 0.28, p-
value = 0.597).
A
B
C
163
Appendix S4.3: Table presenting for each site, sample size (“N”), geographical coordinates (“X”
and “Y”, extended Lambert II), mean expected heterozygosity over all loci (“He”), mean allelic
richness computed using a rarefaction approach after exclusion of small populations (n<14) over
all loci (“Ar”), mean Fis over all loci (“Fis”), mean Fst over all loci (“Fst”), p-value for Hardy-
Weinberg equilibrium (“HW”; a significant p-value indicates departure from equilibrium), mean
body length in mm and its standard deviation (“Length (+SD)”), and distance to the river mouth
(DM) in meters.
Site N X Y He Ar Fis Fst HW Length (+SD) DM
NE1 33 450086 1763239 0.70 6.64 0.08 0.07 0.000 146.0 27.7 8733
NE2 31 450693 1762834 0.69 6.61 0.04 0.06 0.267 151.4 29.9 7972
NE3 31 451002 1762645 0.69 6.65 0.12 0.06 0.004 163.8 29.7 7574
NE4 72 451266 1762411 0.67 6.72 0.07 0.04 0.001 154.1 22.1 7251
NE5 34 451654 1752079 0.72 7.63 0.08 0.04 0.000 148.6 33.6 6663
NE6 60 452526 1761557 0.66 7.02 0.07 0.04 0.000 156.1 21.8 5652
NE7 53 453493 1760779 0.64 6.81 0.02 0.04 0.184 153.4 31.6 4278
NE8 65 453482 1760043 0.60 6.32 0.05 0.05 0.004 141.6 28.9 3467
NE9 61 453438 1759580 0.56 5.78 0.06 0.06 0.032 149.7 31.8 3012
NE10 54 454200 1758023 0.50 4.71 0.00 0.08 0.352 158.4 30.8 914
NE11 19 454463 1757327 0.47 4.49 0.02 0.09 0.653 148.1 24.4 120
RU1 28 448609 1764789 0.65 5.35 0.15 0.13 0.023 157.6 21.9 11203
RU3 5 448965 1763495 0.66 - 0.13 0.08 0.024 159.2 22.8 10288
RU4 34 449017 1763627 0.72 5.79 0.06 0.07 0.074 156.9 28.3 9919
RU6 10 450687 1762692 0.74 - 0.03 0.08 0.654 157.4 24.9 8088
RU7 18 451132 1762779 0.63 6.08 0.08 0.08 0.284 150.9 60.1 7777
RU9 14 451576 1762593 0.66 - 0.01 0.06 0.535 163.1 31.3 7525
RU10 26 451638 1761843 0.61 4.91 0.06 0.07 0.256 146.4 25.8 6750
RU11 33 452762 1761782 0.55 4.63 0.02 0.08 0.883 141.2 37.4 5804
RU12 14 453659 1760889 0.74 - -0.06 0.06 1.000 166.4 33.0 4426
RU16 20 453432 1759173 0.72 6.72 0.04 0.12 0.360 127.2 59.0 2752
SOU 48 - - 0.84 9.39 0.04 0.10 0.12 - - -
AVERAGE - - 0.65 6.23 0.05 0.07 0.266 152.3 31.3
164
Appendix S4.4: List of microsatellites used, with their basic statistics (Na: number of alleles; Ar:
mean allelic richness; He: expected heterozygosity; Ho: observed heterozygosity; Fis: fixation
index; Hw: deviation from Hardy-Weinberg equilibrium (p-value); Fst: overall Fst), and
references.
Locus Na Ar He Ho Fis HW Fst Reference
BS131 20 5.27 0.70 0.67 0.05 0.00 0.14 (Estoup et al. 1998)
One9 9 4.03 0.62 0.65 0.00 0.15 0.13 (Scribner et al. 1996)
Sfo1 17 5.28 0.71 0.74 -0.01 0.09 0.12 (Angers and Bernatchez 1996)
Ssa064 26 6.09 0.75 0.72 0.06 0.00 0.09 Estoup et al. 1998
Ssa103 20 4.92 0.65 0.65 0.04 0.04 0.15 (Skaala et al. 2004)
Ssa417 16 4.31 0.66 0.65 0.04 0.09 0.08 (Cairney et al. 2000)
SsosL311 25 5.16 0.69 0.68 0.05 0.00 0.10 (Slettan et al. 1995)
SsoSL438 10 2.78 0.42 0.43 0.02 0.09 0.16 Slettan et al. 1995
T3-13 19 5.16 0.69 0.72 -0.02 0.44 0.12 Estoup et al. 1998
Ots515NWFSC 3 1.57 0.14 0.15 -0.02 NA 0.02 (Naish and Park 2002)
Ssa-60NVH 26 5.65 0.74 0.74 0.02 0.00 0.10 (Grimholt et al. 2002)
Ssa-TAP2a 4 3.45 0.60 0.63 -0.03 0.59 0.10 Grimholt et al. 2002
Ssa-UBA 13 4.01 0.63 0.36 0.46 NA 0.08 Grimholt et al. 2002
Ssa14 5 1.78 0.19 0.17 0.07 NA 0.27 (McConnell et al. 1995)
Ssa85DU 9 3.72 0.52 0.50 0.07 0.23 0.15 (O‘Reilly et al. 1996)
SsoSL417 17 4.40 0.66 0.63 0.06 0.01 0.08 Slettan et al. 1995
165
Appendix S4.5: STRUCTURE barplot representing the assignment probability of individuals that
were confidently assigned to the captive-bred cluster at the first level of the hierarchy (Q-value >
0.7).
All individuals from the Soueich trout farm (SOU) were exclusively assigned to one subcluster (in
black), whereas the other individuals from the river were assigned to a second subcluster (in gray).
166
Appendix S4.6. Results from models testing the relationship between distance to the river mouth,
level of admixture with captive-bred trout and the interaction between these two variables (along
with the intercept) with allelic richness averaged at the site level and Fst calculated at each site
(see Figure 2 in main text). The table provides beta weights (β) along with standard errors (SE)
and p-values (P). *: p-value < 0.05; **: p-value < 0.01.
Allelic richness
Fst
β SE P
β SE P
(Intercept) 0.262 0.218 0.251
-0.157 0.182 0.401
Distance to the river mouth 0.330 0.232 0.179
-0.355 0.200 0.094
Level of admixture (log) -0.313 0.253 0.238
0.792 0.201 0.001**
Distance to the river mouth * Level of
admixture (log) -0.557 0.184 0.010**
0.298 0.154 0.070
167
Appendix S4.7: Interpretation of first-order interactions and of additive terms when predictors are
involved in a first-order interaction.
In the equation Z ~ b0 + b1X + b2Y + b3XY, the regression coefficient b3 (interaction) represents
the amount of change in the regression of Z on X for a one unit change in Y. As the equation can
be reformulated as Z ~ b0 + (b1 + b3Y)X + b2Y (respectively Z ~ b0 + (b2 + b3X)Y + b1X ), the
regression coefficient b1 (respectively b2) represents the effect of the additive term X (respectively
Y) when the other predictor is set to 0.
For instance, in the model presented in Appendix S6 (AR ~ DM * ADM, with AR standing for the
standardized allelic richness, DM for the distance to the mouth and ADM for log-transformed
admixture level), the beta coefficient associated with the additive term DM (β = 0.330)
corresponds to the influence of DM on AR (upstream increase in allelic richness, that is,
downstream decrease in genetic diversity) for an intermediate level of admixture ADM (that is,
ADM = 0, as all variables were standardized). This specific additive effect of DM is also
illustrated in Figure 2 (main text) in the form of the line corresponding to ―medium admixture‖. In
this fan-representation of the interaction, lines corresponding to ―low‖ and ―high admixture‖ stand
for the specific effects of DM at null and maximum levels of admixture, respectively: Figure 2
shows an upstream increase in allelic richness for null to moderate levels of admixture, but an
opposite trend for high levels of admixture.
Note that an additive regression coefficient may also be considered as the average effect of one
predictor across all values of another variable (Aiken et al. 1991).
168
Appendix S4.8. Three-dimensional plots of (A) individual dispersal propensity and (B) individual
dispersal distance against level of admixture with captive-bred trout and body length in mm (see
Table 1 in main text).
169
Appendix S4.9: Inferring dispersal from full-sibs families
On top of inferring dispersal with GENECLASS2, we additionally inferred dispersal events using
family reconstruction (Berry et al. 2004; Kanno et al. 2011). More specifically, we reconstructed
full-sib families in order to determine whether admixture measured at the family level affected
dispersal traits.
Method
Full-sib families of brown trout were reconstructed using the full-likelihood approach
implemented in COLONY2 (Jones and Wang 2010). COLONY2 implements full-pedigree
likelihood methods, i.e. with likelihood considered over the entire pedigree, to infer sibship among
individuals. We assumed that both sexes are polygamous and we allowed for possible inbreeding.
All individuals were considered as offspring and we defined no a priori candidate parental
genotypes (neither males nor females). Allele frequencies were directly determined from the
genetic dataset using COLONY2. Only the full-sib families with associated inclusion probability
higher than 95 % were retained for further analyses. To ensure the strength of family
reconstruction, we performed two COLONY runs with the same parameters as mentioned above,
and only retained congruent families between both runs.
We then estimated the family dispersal probability, by isolating families clustered on a single site
(suggesting no dispersal) and those spread on several sites (suggesting dispersal). Family dispersal
probability values were obtained by transforming the binomial response variable (i.e. dispersal vs.
non dispersal) with the R function predict(). For families spread over several sites, we measured
the family dispersal distance, taking the river distance between the farthest sites containing
members of a full-sib family.
We then tested the effects of admixture with captive-bred trout (i.e. mean level of admixture at the
family level), sex (i.e. mean percentage of males in each full sib family), and body length (i.e.
mean body length at the family level) on both family dispersal probability and family dispersal
distance. We used Generalized Linear Models (GLM, with Gaussian or Binomial error terms
depending on the response variable) with all continuous variables standardized. We included the
quadratic term for admixture level to test for potential non-linear relationships. As the number of
dispersed families was low (n = 39), we did not test for interaction terms. For each dependent
variable (family dispersal probability and family dispersal distance), we then used a model
170
selection procedure based on the Akaike Information Criteria (AIC) to identify the most
parsimonious set of predictors.
Results
Overall, the 715 fish were assigned to 519 full-sib families with probability higher than 95 %. On
average, reconstructed full-sib families were composed of 1.38 individuals (ranging from 1 to 9).
Among them, 404 individuals were found to be the only members of their families (i.e. singleton).
The other 311 individuals were distributed over 115 families of at least two full sibs, with an
average family size of 2.73 individuals (Figure S9-1A).
Among the 115 families of at least two individuals, 39 were found to be dispersed over at least two
sites (and up to three sites). The distances covered by these families ranged from 367.7 to 7131.8
m, with a median of 1482 m (Figure S9-1B). The other 76 families had all their members on one
unique site. The size of the families was not correlated to the distance covered by the family (r = -
0.020, d.f. = 113, p = 0.82).
Figure S9-1. (A) Histogram of full sibling family sizes, and (B) histogram of distances covered by full
sibling families. Black vertical lines correspond to the median
The best model retained to explain family dispersal probability only comprised simple and
quadratic terms of genetic admixture with captive-bred trout (Table S9-1A). As shown in Figure
S9-2A, families composed either of individuals strongly assigned to the wild cluster or to the
captive-bred cluster are more likely to be spread over more than one site than families composed
of individuals with mixed assignments. Similarly, the model retained to explain family dispersal
distance only comprised simple and quadratic terms of genetic admixture (Table S9-1B); families
with either a low or high assignment probability to the captive-bred cluster are more likely to be
171
spread over longer distances than families composed of individuals with mixed assignments
(Figure S9-2B).
A B
Family dispersal probability Family dispersal distance
β SE P β SE P
Intercept 0.154 0.075 0.043 -0.272 0.163 0.098
Sex / /
Body length / /
Admixture -0.385 0.095 < 0.001*** -0.610 0.206 0.004 **
Admixture 2 0.187 0.063 0.004** 0.275 0.137 0.048 *
Table S9-1. Results from the final models retained for testing the role of sex, body length and admixture
(along with the intercept) on the family dispersal probability (A) and the family dispersal distance (B).
Variables that were not retained in the final model are indicated by a slash bar. The table provides beta
weights (β) along with standard errors (SE) and p values (P). *: p value < 0.05; **: p value < 0.01.
Figure S9-2. (A) Observed (dots) and predicted (line) full sibling family dispersal probability against the
mean level of admixture with captive-bred trout for each family; (B) Observed (dots) and predicted (line)
distances covered by full sibling families against the mean level of admixture with captive-bred trout for
each family.
Removing full sibling families of size two
In salmonids, even in cases of large numbers of microsatellites, Colony has been shown to infer
sibships of size two with low accuracy (Garza et al. 2014); we thus decided to perform the same
analyses but only keeping the families composed of at least 3 siblings. Among the 43 families of at
least three individuals, 17 were found to be dispersed over at least two sites (and up to three sites).
We used the same statistical framework as described above.
172
Although statistical power was much lower due to the lower sample size, we found very similar
patterns using this restricted dataset (Table S9-2, Figure S9-3). There was still a tendency for a
quadratic relationship between dispersal probability and level of admixture (Figure S9-3A) and a
negative relationship between dispersal distance and admixture (Figure S9-3B). Dispersal distance
was also negatively to body size (Table S9-2), as found when using the first migrant approach.
Family dispersal probability Family dispersal distance
B SE P B SE P
Intercept
0.178 0.140 0.212
<0.001 0.142 1.000
Sex
/ / /
/ / /
Body length
/ / /
-0.257 0.144 0.082
Admixture
-0.359 0.129 0.008**
-0.332 0.144 0.027*
Admixture2 0.223 0.124 0.081 / / /
Table S9-2. Results from the models retained for testing the role of admixture (along with the intercept) on
the family dispersal probability (A) and the family dispersal distance (B), with only families of at least 3
individuals. The table provides beta weights (β) along with standard errors (SE) and p values (P).
Figure S9-3. (A) Observed (dots) and predicted (line) dispersal probability of full sibling families of at
least 3 individuals against the mean level of admixture with captive-bred trout; (B) Observed (dots) and
predicted (line) distances covered by full sibling families of at least 3 individuals against the mean level of
admixture with captive-bred trout.
Admixture rate Admixture rate
173
References for supplementary material of chapter IV.
Aiken LS, West SG, Reno RR (1991) Multiple regression: testing and interpreting interactions. Sage
Publications, Newbury Park, Calif
Angers B, Bernatchez L (1996) Usefulness of heterologous microsatellites obtained from brook charr,
Salvelinus fontinalis Mitchill, in other Salvelinus species. Mol Ecol 5:317–319.
Berry O, Tocher MD, Sarre SD (2004) Can assignment tests measure dispersal? Mol Ecol 13:551–561.
Cairney M, Taggart JB, HOyheim B (2000) Characterization of microsatellite and minisatellite loci in
Atlantic salmon (Salmo salar L.) and cross-species amplification in other salmonids. Mol Ecol
9:2175–2178.
Estoup A, Rousset F, Michalakis Y, et al (1998) Comparative analysis of microsatellite and allozyme
markers: a case study investigating microgeographic differentiation in brown trout (Salmo trutta).
Mol Ecol 7:339–353.
Garza JC, Gilbert-Horvath EA, Spence BC, et al (2014) Population structure of steelhead in coastal
California. Trans Am Fish Soc 143:134–152.
Grimholt U, Drabløs F, Jørgensen S, et al (2002) The major histocompatibility class I locus in Atlantic
salmon (Salmo salar L.): polymorphism, linkage analysis and protein modelling. Immunogenetics
54:570–581.
Jones OR, Wang J (2010) COLONY: a program for parentage and sibship inference from multilocus
genotype data. Mol Ecol Resour 10:551–555.
Kanno Y, Vokoun JC, Letcher BH (2011) Sibship reconstruction for inferring mating systems, dispersal
and effective population size in headwater brook trout (Salvelinus fontinalis) populations. Conserv
Genet 12:619–628.
McConnell S, Hamilton L, Morris D, et al (1995) Isolation of salmonid microsatellite loci and their
application to the population genetics of Canadian east coast stocks of Atlantic salmon.
Aquaculture 137:19–30.
Naish KA, Park LK (2002) Linkage relationships for 35 new microsatellite loci in chinook salmon
Oncorhynchus tshawytscha. Anim Genet 33:316–318.
O‘Reilly PT, Hamilton LC, McConnell SK, Wright JM (1996) Rapid analysis of genetic variation in
Atlantic salmon (Salmo salar) by PCR multiplexing of dinucleotide and tetranucleotide
microsatellites. Can J Fish Aquat Sci 53:2292–2298.
Scribner KT, Gust JR, Fields RL (1996) Isolation and characterization of novel salmon microsatellite loci:
cross-species amplification and population genetic applications. Can J Fish Aquat Sci 53:833–841.
Skaala Ø, Høyheim B, Glover K, Dahle G (2004) Microsatellite analysis in domesticated and wild Atlantic
salmon (Salmo salar L.): allelic diversity and identification of individuals. Aquaculture 240:131–
143.
Slettan A, Olsaker I, Lie ø (1995) Atlantic salmon, Salmo salar , microsatellites at the SSOSL25, SSOSL85,
SSOSL311, SSOSL417 loci. Anim Genet 26:281–282.
174
Chapter V. Patterns of brown trout (Salmo trutta) movements
and underlying drivers in replicated riverscapes: an
approach combining isotopic and genetic markers
Saint-Pé Keoni1*
, Laurence Tissot3, Nicolas Poulet
4, Julien Cucherousset
2, Jérôme G. Prunier
1, and
Simon Blanchet1,2*
1 Centre National de la Recherche Scientifique (CNRS), Université Paul Sabatier (UPS); Station
d‘Ecologie Théorique et Expérimentale, UMR 5321, F-09200 Moulis, France
2 CNRS, UPS, École Nationale de Formation Agronomique (ENFA) ; UMR 5174 EDB
(Laboratoire Évolution & Diversité Biologique), 118 route de Narbonne, F-31062 Toulouse cedex
4, France
3 EDF Recherche et Développement, Laboratoire National d‘Hydraulique et Environnement, 6
quai Watier, 78401 Chatou Cedex, France
4 Pôle écohydraulique AFB-IMT, allée du Pr Camille Soula, 31400 Toulouse France
Key words: Movement, brown trout, SNPs, isotopes, machine learning, individual assignment.
175
Summary
In wild populations, determinants of movements lie in individual, environmental and populational
characteristics. Assessing how these three components interact in shaping dispersal patterns is thus
of high interest for both fundamental and applied issues. Using an original assignment method
combining isotopic and genetic markers, we investigated the effects of individual phenotypes,
environmental and populational indicators, and of connectivity features in shaping movement
patterns in brown trout (Salmo trutta) populations living in three replicated riverscapes. To do so,
we genotyped 1140 brown trout and collected fin tissue d15N and d13C values to identify
residents and dispersers through individual assignment tests from random forest algorithms. We
tested whether individual characteristics (morphology, heterozygosity…) are linked to dispersal
propensity and movement characteristics. Then, we investigated –within each riverscape- which
populations are more likely to receive immigrants or contrary to be emigrated from, and what are
the population characteristics (biotic and abiotic) sustaining these dynamics. Finally, we tested
whether individual exchanges between pairs of sites are linked to connectivity, and to
environmental and populational similarity between sites. We found that individual size was a main
driver of movement patterns. At the population level, we found that the position of the site within
riverscapes, and habitat availability influenced the number of immigrants a site is likely to receive,
and these relations varied between river basins. Finally, we found that sites that are directly
connected by the water flow, and those that are similar in terms of elevation and stream order
exchanged more migrants. These findings have implications for conservation, as identifying
individual traits and environmental characteristics which may favor movements can be of use in
defining populations or habitats of high conservation interest, and thus prioritizing conservation
efforts.
176
Introduction
Movement is a fundamental process for population dynamics and evolution, gene flow, and
species distribution (Clobert, 2012). It is now common knowledge that important population
processes depend on individual movement, almost as much as on fluctuations in population
density (Patterson et al., 2008). This has widely been shown for various types of ecosystems, with
examples on metapopulation ecology (Hanski, 1998), spread of diseases and invasive species (Kot
et al., 1996), or home-range shifts (Simmons and Thomas, 2004), and thus has great implications
in terms of management. Because movement is a determinant mechanism allowing species to
respond to changing environmental conditions (Brooker et al., 2007; Haddad et al., 2015; Ims,
1995), understanding movement patterns and their causes appears crucial to optimize conservation
efforts and predict consequences of environmental changes (Caplat et al., 2016). However,
interactions between life history, physiology, behavior and habitat make individual movement an
exceptionally complex phenomenon (Patterson et al., 2008). Movements are characterized by three
distinct steps (Clobert, 2001; Clobert et al., 2009): emigration (i.e. leaving a location), transience
(i.e. the movement per se, during which the displacement of the individual takes place), and
immigration (i.e. the settlement in the destination location). Although an important component of
movements is random (Cain et al., 2000; Hawkes, 2009), it is now acknowledged that many
animals do not (only) move randomly as it is very commonly assumed in numerous models of
spatial population dynamics and species distribution. Indeed, various studies showed that
individuals move due to particular determinants which can act on all stages of movement: decision
to leave, transience, and choice of destination location (Clobert, 2012; Cote et al., 2010). These
factors driving movements have thus been a field of interest for many studies (Bowler and Benton,
2005; Ronce, 2007).
Movement implies costs and risks for individuals, at all stages of the process (Bonte et al. 2012).
At a local scale, both biotic and abiotic features may indeed trigger emigration and/or motivate
immigration, depending on individuals‘ expected fitness (e.g., Holt and Barfield, 2015). Dispersal
patterns were for instance found to be linked to spatio-temporal changes in population density
(McPeek and Holt, 1992), intraspecific competition and inbreeding risk (Branch, 1975; Szulkin
and Sheldon, 2008; Johnson and Gaines, 1990), predation risk avoidance (Matthysen, 2012) or
local habitat quality (Lin and Batzli, 2001; Lurz et al., 1997). Examples of movement patterns
177
shaped by these mechanisms are numerous, for instance increased numbers of post-natal dispersers
(Monard and Duncan, 1996) and sex-biased dispersal (Cockburn et al., 1985; Pusey, 1987). At a
regional scale, transient individuals may also struggle to access or find suitable habitats because of
limited dispersal capacities (isolation-by-distance, IBD) or an increase in mortality risk, as they
may for instance become exposed to predators or to physiologically harsh conditions (isolation-by-
resistance, IBR; Edelaar et al., 2008; Bonte et al., 2012). For instance, a classical example is that
of river dams, which impede movement, and increase mortality during passage of these obstacles
(Gouraud et al., 2008). Paz-Vinas et al., (2015) also showed that individuals were more likely to
move between two locations which are flow connected, i.e. between which water flows from the
upstream deme to the downstream deme (Peterson et al., 2013).
These costs can be reduced through a plastic response of individuals or through the selection of
specialized traits in terms of morphology, physiology or behavior (Clobert et al., 2009, Cote et al.
2010) . These specializations have been investigated for several decades now, and it is shown that
there are both quantitative and qualitative differences between dispersing and resident individuals
(Myers and Krebs, 1971), which are shaped by specializations needed to increase movement
success (Clobert et al., 2009). For instance, some studies showed that dispersers are larger, fatter
and have higher locomotor and feeding activities (O‘Riain et al., 1996). Qualitative (behavioral)
traits were also shown to play a key role in movements having important ecological consequences,
with interesting examples concerning animal personality. Examples are numerous, where
dispersers and residents have shown differences in activity, boldness, aggressiveness, social
behavior, and mating decisions (Cote and Clobert, 2007; Dingemanse et al., 2003; O‘Riain et al.,
1996). Finally, in a study on mountain goats (Shafer et al., 2011), individuals exhibiting lower
heterozygosity were found to disperse more than those with higher heterozygosity, and the authors
suggest that higher fitness can be associated with more residency as heterozygosity is associated
with fitness. Beyond phenotypic specialization, dispersing individuals may also increase their
expected fitness by actively looking after environmental conditions that best match their
phenotype, a hypothesis named habitat matching (Edelaar et al., 2008; Jacob et al., 2015) and
resulting in patterns of isolation-by-environment (IBE, Cayuela et al., 2018; Wang and Bradburd,
2014), as individuals tend to settle in habitats similar to the one they come from (and are adapted
to).
178
In riverine systems in particular, the shape of the network, as well as hydrological features (slope,
flow, speed of flow, obstacles etc…) were shown to be determinant in shaping patterns of
movements (Carrara et al., 2014; Hitt and Angermeier, 2008; Labonne et al., 2008; Paz-Vinas et
al., 2015). For instance, upstream reaches, which are more isolated and provide smaller habitat
availability were shown both theoretically and empirically to have much less immigrants than
larger ones (Tonkin et al., 2018). The position of a deme within the riverscape is also crucial in
shaping these patterns, as more central ones (i.e those which are physically easier to access to from
surrounding ones, Paz-Vinas and Blanchet, 2015) were also shown to receive more immigrants
than more isolated ones (Altermatt and Fronhofer, 2018). Moreover, anthropogenic pressures on
these systems were also shown to modify dispersal patterns, either through the environment
alteration (fragmentation, Brooks et al., 2018; Sondermann et al., 2015; pollution and flow
reduction, Souza et al., 2008), or through the introduction of exotic phenotypes (for instance by
stocking individuals whose genetics and phenotypic background is different; Jonsson et al., 1991;
Saint-Pé et al., 2018). Altogether, this suggests that complex links between environment and
individual characteristics are expected in river systems in shaping dispersal patterns. Describing
and understanding patterns of connectivity and movements within river systems, and to which
extent patterns and processes are similar and generalizable across river systems, are keys to
investigate ecological processes in rivers, and to improve river management and conservation
efforts (Pringle, 2001).
Here, we aimed at describing riverscape-scale movements in a widely distributed headwater
stream fish (the brown trout, Salmo trutta) and at unraveling the links between environment and
individual characteristics in replicated river systems. To that aim, we used an original assignment
method combining isotopic and genetic markers in a machine learning algorithm to identify
migrant/resident individuals and to determine the population of origin of immigrants (Berry et al.,
2004; Manel et al., 2005). These markers have been used successfully for assignment (Cornuet et
al., 1999; Paetkau et al., 1995; Hobson, 1999; Martin et al., 2013), and using them in combination,
notably isotopic and genetic ones, was recently shown to be highly accurate (Clegg et al., 2003 ;
Saint-Pé et al. in prep). More specifically, we aimed at identifying (i) individual specializations of
individuals correlated with movement propensity and distance travelled (ii) environmental
determinants influencing individual movement success including both transience and immigration
179
(IBD, IBR and IBE), and (iii) environmental/positional determinants of attractivity of a site,
estimated by the proportion of immigrants detected at each site. First, we expected moving
individuals to show higher body condition/length as they are more likely to afford the costs of
movement (O‘Riain et al., 1996), and that individuals exhibiting higher heterozygosity should be
less mobile, as this was shown in Shafer et al., (2011). Further, we expected that IBD, IBR and
IBE to shape patterns of movements between populations. More precisely, we expected that
populations which are farther in terms of riparian distance, which are not flow connected, which
are separated by more obstacles, and which are more dissimilar in terms of environmental
conditions (temperature for instance) should exchange less migrants than well connected, close
and environmentally similar populations. Finally, we expected that populations which are more
central, and offering larger habitat availability, which are generally in lower reaches of the river
catchment would be the most attractive within the network, therefore receiving more immigrants,
as should locations in which environmental conditions are more favorable (lower temperatures for
instance). We performed that work in three independent river systems in order to test whether
patterns and processes underlying movement patterns in brown trout are similar across different
ecological contexts, or on the contrary context-dependent. This last objective has strong
implications for conservation strategies as it will help identifying whether conservation programs
related to landscape connectivity should be context-dependent or not.
Methods
Field sampling
We sampled three independent French river basins (Figure 1) in which the fish community is
dominated by brown trout (Salmo trutta), the Aude River (Mediterranean outlet), the One River
(Atlantic outlet), and the Seuge River (Atlantic outlet). Brown trout were sampled using a single-
pass electric-fishing approach from 33 sites in total, with an aim of 30 individuals per site. We
focused on small individuals in order to focus mainly on one age class (1 year-old). In total, we
captured 1325 individuals (453 for the Aude River, 495 for the One river and 377 for the Seuge
River), with sample sizes ranging from 6 to 30. Each individual was measured (total length in mm;
Table 1) and a pelvic fin clip was taken for genetic and isotopic analyses. All individuals were
released alive to their original sampling site.
180
Figure 5.1. Maps of the study sites, representing the three river basins sampled (the Aude, the One and the
Seuge Rivers). Black dots represent sampling locations, arrows at the end of the river lines represent
direction of the flow.
AUDE ONE SEUGE
Site N Body length (±SE) in mm Site N Body length (±SE) in mm Site N Body length (±SE) in mm
Agu-Sou 27 157(±42) Cou-Lar 30 111(±26) Ber-Ben 11 114(±28) Aig-Pou 30 139(±33) Lab-Cas 17 112(±23) Bui-Pie 6 96(±27) Aig-Sou 30 143(±38) Nga-Jur 30 152(±40) Cla-Tou 9 105(±55) Art-Lau 30 138(±38) Nga-Mar 30 135(±18) Lav-Riv 15 138(±51) Aud-Car 30 166(±34) Nga-Vga 30 127(±35) Lav-Suc 30 122(±31) Aud-Far 30 165(±33) Noo-Ast 30 135(±26) Pon-Amo 30 127(±26) Aud-Fou 30 135(±25) Noo-Cas 30 144(±18) Pon-Bom 30 133(±32) Aud-Nen 30 123(±54) Noo-Esp 30 129(±38) Pon-Cha 30 148(±29) Aud-Puy 30 121(±31) Noo-Lac 30 109(±26) Pon-Tis 30 129(±18) Aud-Ser 30 156(±22) Noo-Sav 30 111(±63) Ser-Sau 6 143(±28) Bai-Pas 14 153(±28) Noo-Tre 30 157(±33) Seu-Cch 30 110(±21) Bru-Mij 30 127(±26) Noo-Voo 30 138(±30) Seu-Cou 30 134(±27) Bru-Uss 30 151(±21) Nou-Boo 28 173(±43) Seu-Cro 30 165(±33) Cam-Sau 14 118(±46) Nou-Cir 30 134(±42) Seu-Fag 30 126(±32) Que-Mas 29 131(±36) Nou-May 30 155(±34) Seu-Rod 30 157(±38) Que-Ria 30 158(±37) Nou-Spo 30 140(±38) Seu-Sau 30 138(±19) Roq-Sau 9 154(±25) One-Bag 30 147(±28)
Total 453 143(±37) 495 136(±38) 377 134(±34)
Table 5.1: For each river, site code, number N of individuals sampled in each site, and mean body length in
mm (± standard error).
181
Genotyping
Individual multilocus genotypes were obtained for 162 SNPs markers (Saint-Pé et al., 2019) on the
1325 individuals (453 for the Aude River, 495 for the One river and 377 for the Seuge River),
using KASPAR technology performed by LGC Genomics (Smith and Maughan, 2015). All
polymorphic markers were kept regardless of selection and linkage disequilibrium, as these
processes can be exploited for investigating population structure and increasing assignment power
(Morin et al., 2009; Nielsen et al., 2009; Waples and Gaggiotti, 2006). For further analyses, we
discarded individuals for which more than one third of SNPs were not genotyped, leaving us with
1314 individuals (449 in the Aude River, 375 in the Seuge River, 490 in the One River).
As stocking with captive-bred individuals may heavily impact inferred movement patterns,
notably in brown trout (Saint-Pé et al., 2018), we chose to focus on wild individuals only. We thus
removed individuals which were assigned to captive-bred populations with a probability higher
than 0.3, using STRUCTURE (Pritchard et al., 2000) and following the same procedure as in
Saint-Pé et al. (2018). This concerned 48 individuals in the Aude River and 126 in the One River,
the Seuge basin showed only very slight traces of hatchery ancestry (all individuals assigned at
less than 0.3 to the captive-bred cluster). The ―wild‖ genetic dataset comprised 1140 individuals
(401 individuals in the Aude basin, 364 in the One, and 375 in the Seuge), and only these
individuals were kept for subsequent analyses.
Carbon and Nitrogen stable isotopes
For all 1140 wild individuals and successfully genotyped, fin samples were dried at 60 °C for 72
hours, and then loaded into 1.5mL Eppendorf tubes, labeled and parafilmed. These samples were
analyzed at Cornell University (COIL, Ithaca NY), using an elemental analyzer plumbed into a
Finnigan MAT Delta Plus isotope ratio mass spectrometer to measure δ13
C and δ15
N.
Inference of movements
Assignment was performed on all 1140 remaining individuals (i.e. wild and successfully
genotyped). For each river basin, using the combination of genetic and stable isotopes of Carbon
and Nitrogen, we calculated individual assignment probability to each site using a classification
182
method based on the Random Forest RF algorithm, in which the specified model aims at
predicting the ―site‖ from both genetic and isotopic data (Saint-Pé et al., in prep). The number of
trees for the classification tested ranged from 1 to 6000, and the model retained made sure the
number of trees was sufficient for the out of bag OOB error rate (classification to sites different
than capture site) to be stable and lowest. The number of splits of the variables (―mtry‖ parameter)
was also chosen to have the lowest OOB error rate (default value, then default*2, then number of
variables). We then used a dedicated permutation procedure to filter out dubious individuals and
avoid type 1 errors, as described in Saint-Pé et al. (in prep). In short, we permuted the site (capture
site) 1000 times, and performed RF classification on all the 1000 permuted datasets, with the same
number of trees and dataset splits retained in the first step. From this, we calculated for each
individual the 95th quantile of maximum assignment over the 1000 runs. Individuals were
considered as ―resident‖ if the probability of assignment was highest for their capture site, and
considered as ―migrant‖ if the probability of assignment was highest for another site and this
probability was higher than the 95th
quantile of maximum assignment over the 1000 permutations.
Origins of individuals assigned to a site different than their capture site, but not with a probability
higher than the quantile of the permutated analyses were considered as ―unknown‖ (thus avoiding
type 1 errors) and not considered further into analyses.
Determinants of movement at the individual scale
For individuals considered as migrants, we calculated the distances travelled from inferred site of
origin to site of capture using the STARS ArcGis package (Peterson and Hoef, 2014). Migration
propensity (binomial response: migrant or resident) and distances travelled by migrants were then
used as response variables in subsequent analyses in order to identify individual determinants of
movements.
We then measured/calculated individual characteristics that served as explanatory variables for
migration propensity and migration distance. First, body length of each individual was measured
in mm. Then we calculated Fulton‘s K body condition index (body weight/body length3, Froese,
2006). Further, for morphometrics measures, each individual was placed on a polystyrene board
with a ruler and a color sample, and was photographed with a digital camera, at the same
magnification. First, using ImageJ, we measured a total of 12 traits having been related to
183
movement ability functions in fish (Albouy et al., 2011; Bracciali et al., 2016, Appendix S1). A
Principal Component Analysis (PCA) was performed on these traits, and we used the first two
axes of this PCA as explanatory responses because they explained more than 75% of variance
(Appendix S2). The first axis represents the degree to which an individual is longiform, while the
second axis (towards smaller values) represents the height and width of the back half of the body
in comparison to the front. Finally, we measured individual heterozygosity as the number of loci
on which an individual is heterozygote/number of loci.
We then tested movement propensity (migrant vs. resident) against the previously detailed
explanatory variables using a binomial generalized linear model (glm R function), and with the
river basin as an interaction term with all variables. We then performed a Type II ANOVA using
the Anova function from the car R package to determine the terms in the model used (including
interaction) which significantly participated in partitioning variance of the response variable.
Similarly, we tested migration distance (after log-transformation as it was not normally
distributed, data not shown) against the explanatory variables described above, using a Gaussian
generalized linear model (glm R function), also with an interaction with the river basin with all
simple terms. Again, we performed a Type II AVONA using the Anova function from the car R
package.
Environmental and populational determinants of emigration and immigration
For each site, we calculated the proportion of immigrants as the number of migrants divided by the
total number of individuals at the site level, and used that proportion as the response variable in
subsequent analyses. We chose not to consider the possible number of emigrants inferred from
assignment, as even if it is possible to know how many individuals were migrants and came from a
specific site in our sample, we suggest that the majority of emigrants most likely moved to a
location which we did not sample, therefore, this estimation should be highly erroneous because
our sampling design was not exhaustive.
Then we retrieved for each site the environmental variables that were used as explanatory
variables. First, the hierarchical classification of stream (Stralher index, Shreve, 1966) to which
the sites belong; higher values indicating lower downstream stretches with lower branching
184
complexity and thus closer to the river mouth. Second, we calculated the centrality index with the
STARS ArcGis package, that quantifies the positional importance of a node within a network (to
which extent a site is connected to all others, Freeman, 1977). Third, we measured the physico-
chemistry features of each site, including altitude, riverine distance to river mouth (calculated with
the STARS ArcGis package), and water quality as measured from a multiparameter probe (water
temperature, dissolved oxygen, and conductivity). For each site, measures of water quality were
taken punctually, twice (once in the morning and once in the afternoon). We performed a PCA on
these variables and we used the first two axes of this PCA as integrative explanatory responses as
they explained more than 75% of variance (Appendix S3A). The first axis represents a temperature
gradient, with higher temperature, and thus lower oxygen, towards high values, while the second is
an altitudinal axis (higher values towards higher altitude and therefore farther from the river
mouth). Finally, we characterized habitat availability by measuring mean depth and width of the
river at each sampling site. Depth and width were averaged across two transects at each site during
which water depth was measured every meter along transects. We performed a PCA to synthetize
these two variables onto a single axis (Appendix S3B). The higher the value on that axis, the
higher the depth and width, and thus the higher the habitat availability.
The proportions of immigrants at each site being typically beta-distributed (Ferrari and Cribari-
Neto, 2004), they were then regressed against these environmental and populational characteristics
using beta regressions (betareg function from the betareg R package). To control for zero
inflation, proportions were transformed using the following equation: (y * (n − 1) + 0.5)/n where
n is the sample size (Smithson and Verkuilen 2006). The effect of river basin was considered in
interaction with each other predictors. As previously, we finally performed a type II ANOVA
(Anova function from the car R package).
Environmental determinants of interdeme movement success
Finally, we assessed if the number of movements between all pairs of sites (response variable) was
related to (i) Flow connectivity (Paz-Vinas et al., 2015: two demes were considered flow-
connected when water can flow from the upstream deme to the downstream deme; flow-
unconnected demes are two demes that share a common confluence downstream but do not share
flow as in Peterson et al., 2013) (ii) Riverine distance (calculated using the the STARS ArcGis
185
package, as mentioned previously) (iii) Number of riverine obstacles (obtained from the French
water agency database ―Référentiel des obstacles à l‘écoulement‖, ROE), and (iv) isolation by
environment parameters (calculated first by computing the distance (pairwise differences) along
the previously described PCA axes, and then by computing pairwise difference in Strahler index
between sites).
To control for the independence of pairwise data, we here considered a mixed-effect model with
Maximum-likelihood-population-effect (MLPE) error structure. MLPE error structures are
designed to account for the non-independence of pairwise data (Clarke et al., 2002; Van Strien et
al., 2012).. MLPE models have been successfully used to identify landscape features that impact
movements (Blair et al., 2013; Peterman et al., 2014; Row et al., 2017). The fixed part of our
model included pairwise landscape matrices (described above), with the river basin as an
interaction term. As we aimed at explaining numbers of movements between pairs of sites, we
considered a MLPE mixed-effect model with a Poisson error distribution. The intercepts and
slopes for the fixed effects were estimated with restricted maximum likelihood (REML, lme4
package in R; Clarke et al., 2002; Van Strien et al., 2012). Finally, we performed a Type II
ANOVA (Anova function from the car R package) on this model.
Results
Description of movements detected
The proportions of individuals classified as residents were 62% in the One River, 72% in the
Seuge River and 84% in the River, whereas the proportions of immigrants (i.e. individuals not
originating from their capture sites) were 12% in the One River, 24% in the Seuge River, and 15%
in the Aude River (Figure 2A). Other individuals were classified as ―unknown‖ and were not
further considered in the analyses. Individuals which were significantly detected as migrants
showed different movement patterns across rivers (Figure 2D; see also appendix S4 for a
cartographic visualization). The distances travelled by these individuals ranged from 585 to 31584
meters, with a median of 7609 meters. The directions of these movements also varied among
rivers (Figure 2B). For instance in the Aude River, there were roughly the same proportions of
movement directions (28% downstream, 33% upstream, and 38% implying first downstream then
upstream movement, Figure 2B), whereas in the One River, most movements occurred upstream
186
(64% vs 34% downstream and only 2% implying both down and upstream movement). Finally,
most movements occurred between two streams of equal Strahler order (44%, 37% and 5% of
movements of between sites belonging to 1, 2 and 3 Strahler orders respectively, whereas all other
movements, between different types of streams were less frequent (>5%) (Figure 2C).
Figure 5.2. Numbers of residents, migrants and non-assigned individuals (A), and for migrants, direction
(B), Strahler of origin and destination sites (C), and density (Y axis) of distances travelled (D).
187
Individual specialization in movement
The Fulton body condition index K in interaction with the river basin, and body length alone were
found as significant predictors of movement propensity (Table 2, ―movement propensity‖). More
precisely, we found that in the One River, individuals with a higher body condition tended to
move less, whereas in the two other rivers (Aude and Seuge), individuals with higher body
condition were more likely to move (Figure 3A; summary of complete model presented in
Appendix S5). Additionally, longer individuals were more likely to move, a finding that holds true
for all river basins (Figure 3B, Appendix S5).
Movement propensity Movement distance
Chisq Df Pr(>Chisq) Chisq Df Pr>Chisq)
Body length 10.958 1 0.001 *** 2.965 1 0.085 .
BV 13.799 2 0.001 ** 178.946 2 <2e-16 ***
MorphAx1 1.443 1 0.23
1.866 1 0.172
MorphAx2 2.261 1 0.133
1.376 1 0.241
FultonK 0.347 1 0.556
0.006 1 0.940
Heterozygosity 2.811 1 0.094 . 0.855 1 0.355
Body length:BV 3.722 2 0.156
7.582 2 0.023 *
BV:MoprphAx1 0.235 2 0.889
0.979 2 0.613
BV:MorphAx2 3.249 2 0.197
3.095 2 0.213
BV:FultonK 9.42 2 0.009 ** 2.699 2 0.259
BV:Heterozygosity 2.097 2 0.35 1.169 2 0.557
Table 5.2: Individual determinants of movements. (A) Correlations between explanatory variables used, (B)
ANOVA of both movement propensity and movement distance models, comprising simple terms and
interactions with the river basin
We further found that body length in interaction with the river basin was the only significant
predictor of distance travelled (Table 2B, ―movement distance‖). Longer individuals travelled over
shorter distances in the Aude and in the Seuge Rivers, whereas in the One River, longer
individuals tended to move on longer distances (Figure 3C; summary of complete model presented
in Appendix S6).
188
Figure 5.3. Predictions for migration propensity according to (A) the interaction between body condition
and river basin, and (B) body length. (C) Predictions for log of distance according to the interaction
between body length and river basin. Confidence intervals (95% in grey).
Environmental determinants of individual movement success
Using MLPE mixed-effect models to explain the number of movements among pairwise
populations, we found that flow connectivity between sites, distance along the altitudinal axis, and
difference in Strahler between sites significantly explained the pairwise number of movements
between sites (Table 4B, summary of complete model presented in Appendix S7). More precisely,
flow connectivity was positively correlated to pairwise movement between pairs of sites (Figure
5A), and negatively correlated with distance along the altitudinal axis and difference in Strahler
index (Figure 5 B and C). This means that two sites which are flow connected, similar in terms of
altitudinal conditions, and on the same type of stream are more likely to exchange individuals.
189
Interestingly, we found no relation between pairwise number of individual movements between
sites and riparian distance (p-value = 0.94), which was one of the most expected relation, nor with
obstacles separating sites (p-value = 0.79).
Chisq Df Pr(>Chisq)
BV 2.080 2 0.354
PairwiseObstacles 0.073 1 0.787
PairwiseFlowConnect 10.235 1 0.001 **
PairwiseDist 0.007 1 0.935
dTempAx1 0.582 1 0.446
dAltAx2 7.392 1 0.007 **
dStrahler 4.604 1 0.032 *
BV:PairwiseObstacles 0.588 2 0.745
BV:PairwiseFlowConnect 5.238 2 0.073 .
BV:PairwiseDist 0.558 2 0.756
BV:dTempAx1 0.909 2 0.635
BV:dAltAx2 3.799 2 0.150
BV:dStrahler 2.984 2 0.225
Table 5.4: Tables corresponding to the pairwise approach models. (A) Correlations between explanatory
matrices used, (B) ANOVA of the model explaining the number of pairwise exchanges between sites,
comprising simple terms and interactions with the river basin
Figure 5.5. Predictors for the number of pairwise movements between sites according to flow connectivity,
difference in the altitudinal axis, and difference in Stralher index. Dotted lines represent confidence
intervals (95%)
Environmental determinants of population attractivity
At the local/population scale, the proportion of immigrants was significantly explained by three
terms in interaction with the river basin: habitat availability, Strahler index and Centrality (Table
3B, ―proportion of immigrants‖). More precisely, the proportion of immigrants was positively
190
related to habitat availability in the One and the Seuge Rivers, meaning that larger habitats tend to
receive more immigrants, whereas the opposite tendency was observed in the Aude River (Figure
4A). The proportion of immigrants was positively related to Strahler index in the Aude and the
One Rivers, whereas this relation was negative in the Seuge River (Figure 4B). Proportion of
immigrants was negatively correlated to centrality in the Aude and in the Seuge Rivers, but
positively correlated to this variable in the One River, highlighting the importance of site position
within the dendritic network (Figure 4C). On the other hand, variables participating to the
temperature axis (i.e. oxygen, temperature, conductivity) and to the altitudinal axis (altitude,
distance to the river mouth) did not explain the proportion of immigrants. Complete model
summaries are presented in appendix S8.
Proportion of immigrants
Chisq DF Pr(>Chisq)
TempAx1 0.125 1 0.723 BV 3.226 2 0.199 AltAx2 0.599 1 0.439 HabAvailAx1 3.208 1 0.073 .
Strahler 0.700 1 0.403 Centrality 0.720 1 0.396 TempAx1:BV 4.357 2 0.113 BV:AltAx2 3.722 2 0.155 BV:HabAvailAx1 6.675 2 0.036 *
BV:Strahler 7.256 2 0.027 * BV:Centrality 6.534 2 0.038 *
Table 5.3: Environmental determinants of population attractivity ANOVA of both proportion of immigrants
and proportion of emigrants at each site, comprising simple terms and interactions with the river basin
191
Figure 5.4. Predictors for the proportion of immigrants at the site level (A) habitat availability, (B)
Strahler index, and (C) centrality
192
Discussion
In this study, we investigated patterns and determinants of movements in three replicated river
basins at different levels (from individual features to site and populational features). We found that
individuals tend to move according to body size mainly, both in terms of propensity and
movement distance. We also found that sites which are most likely to exchange migrants are those
which are connected by water flow, and similar in terms of elevation and stream type. Finally, we
found that according to river basins considered, some types of locations can be more or less likely
to receive immigrants, in particular depending on both habitat quality (habitat size) and
accessibility, notably due to their position in the river network (centrality and Strahler index).
Overall, determinants were in the most cases river-dependent, which shows that patterns and
processes of fish movements are hardly generalizable across riverscapes.
Individual specialization in movement
As suggested by our results, individual size is a major determinant of movements. More
particularly in the case of the Brown trout, it was shown that these dynamics are often the result of
hierarchical interactions between individuals, and exist at various life stages (Bremset and Berg,
1997; Cuinat and Heland, 1979). We found that regardless of the basin, longer individuals were
more likely to move. However, in the Aude and the Seuge Rivers, these larger individuals moved
on shorter distances, whereas in the One River longer individuals moved on longer distances. For
the One River, we can therefore suggest that smaller individuals often disperse further because
more dominant and often larger individuals attain habitats of first choice which they subsequently
defend (Heland, 1980) for instance. Moreover, larger and more dominant fish tend to have a more
fixed, but wider home range than smaller and subdominant individuals (Elliott, 1994; Höjesjö et
al., 2007), although they have better movement capacities. Thus, movement patterns may be very
context dependent, and determined by local trade-offs between remaining resident and moving
which is modulated by body condition and size.
Body condition and metabolism in general have been shown to be determinant in individuals‘
decisions to move (Baines et al., 2015; Niitepõld et al., 2009), as a good condition may generally
enable an individual to afford the cost of movement (Bonte et al., 2012). Accordingly, individuals
193
from the Aude and the Seuge Rivers exhibited a higher probability to have moved when their body
condition index was higher. However, in the One River, the opposite pattern was observed. This
could imply that in this particular basin, movement decisions may be forced instead of chosen: if
their environment or biotic interactions do not enable them to sustain a suitable condition,
individuals may decide to move or be forced to move out by individuals in better body condition
(Cote et al., 2010; Hoset et al., 2011).
Morphometric ratios related to swimming ability were not retained as significant in the models
explaining individual movement. This is relatively surprising, as even at these small spatial
extents, variations in selection pressures would have been expected to be sufficient to induce
morphological differentiation between individuals (Pakkasmaa and Piironen, 2001), however we
suggest that different morphotypes may not be contrasted enough to enable to link them to
differential movement behavior.
Individual heterozygosity was unexpectedly not linked to individual movement either. According
to our predictions, we were expecting that due to the links between this variable and fitness
(Shafer et al., 2011), individuals exhibiting lower heterozygosity should disperse more than those
with higher heterozygosity.
Pairwise determinants of exchange between sites
In particular in the context of habitat fragmentation, movements between patches have been a vast
field of research (Travis et al., 2013). Pairwise approaches are numerous, and generally, they all
show that between sites movements are negatively related to inter patch resistance, i.e. the
difficulty for an individual to move between patches. In rivers, this has also been investigated, for
instance showing that gene flow between patches is negatively related to distance (Paz-Vinas et
al., 2015), and obstacles to physical connectivity (Gouraud et al., 2008). Surprisingly, riparian
distance did not affect the number of movements between two sites, however, other variables
linked to physical and landscape resistance did, i.e. flow connectivity between sites, but also
difference along the altitudinal axis and difference in stream type. This could suggest that
movements of trout may have stronger biological causes than just spatial probability associated
with proximity. On the other hand, movements between sites were not linked to variables related
194
to habitat choice: we found no effect of temperature, however, other variables which were not
tested could be more likely to explain habitat suitability choice in movement decisions between
sites. Finally, the fact that anthropic obstacles did not appear to reduce movements between
locations although this was expected could suggest that in brown trout studies, these obstacles are
often overestimated: depending on water flow in particular, assessing more thoroughly how they
impede movement is a lacking field at these small scales.
Environmental determinants of population attractivity
Environmental factors which influence how a certain deme will contribute to the regional
emigration-immigration dynamics within a river system were shown to be linked to the position of
that deme within the dendritic network (Carrara et al., 2014; Paz-Vinas and Blanchet, 2015), but
also to its size: upstream, smaller and less connected demes were shown to generally produce
emigrants, which immigrate to downstream, more connected and larger demes. This is also very
congruent with the life cycle of the Brown trout: migrations to mate upstream, with subsequent
downstream dispersal (Maisse and Bagliniere, 1990; Ovidio et al., 1998; Roussel and Bardonnet,
2002). In our study, we found that these ―positional‖ features at the site level influenced the
proportion of immigrants: habitat availability, stream type and centrality. Except in the Aude
River, sites with higher habitat availability for instance tended to receive more immigrants,
however, there were many exceptions going against these theoretical expectations. For example in
the Seuge river, higher Strahler index streams (that is, downstream river stretches) tended to
receive less immigrants, and in the Aude and the Seuge Rivers, centrality was also negatively
related to the proportion of immigrants. These counter intuitive exceptions thus highlight the need
for very context-specific studies. Indeed, as different basins vary considerably in terms of
environmental features, habitat use and spatial scales considered for different facets of ecology
may be very different, and thus conclusions must be drawn in regards of these different
possibilities.
We expected other variables, in particular temperature and oxygen but also altitude and distance to
the river mouth to influence these movement choices because we expected that some sites may be
better suited for immigrants. This was surprisingly not observed, as in explaining pairwise
determinants of movement, suggesting that individuals may not decide to immigrate to a location
195
in which environmental conditions are more favorable (i.e. higher oxygen, lower temperature,
etc…), or at least, these choices are negligible compared to those based on the size and position of
a destination site.
Technical considerations and conclusion
Generally, rivers inhabited by Brown trout are all affected to a certain extent by human alteration.
A major perturbation, which in this study we decided to not consider is stocking, however, its
effects are of utmost importance: first, movement inference may strongly be biased by this
practice, human induced displacements may be detected as movements when using assignment
tests. Moreover, in terms of ecology, even if it may be possible to discard captive-bred individuals
from these analyses, they were shown to exhibit different movement behavior (Saint-Pé et al.,
2018), and as they are part of the population and its dynamics, their impact on movement behavior
of wild populations with which they seem to be living with in parapatry mostly remains to be
assessed.
A final however extremely important consideration is the use of data types in combination. By
combining genetics and isotopes to detect movements, and in particular because the aim of this
study is ecological, we suggest that we may be missing important biological insight, as the types
of movements detected by isotopes may differ from that of genetics (Cook et al., 2007). We
therefore call for future studies disentangling these time scale considerations, and in the best case,
confirming that these markers can be used for very different purposes with direct tracking
protocols such as telemetry.
Nevertheless, using a robust assignment method applied to wild individuals, we were able to
identify key drivers of natural movement patterns in different river contexts. We showed that
brown trout movements are driven by complex interactions between individual traits and
environment. Moreover, these drivers are very context dependent, and seem to depend on the
configuration of the riverscape. This has great implications in terms of conservation because it
implies that each river must be studied and managed in its own way. This study opens new insight
into multi-riverscape studies, because the approaches we developed make the study of movements
possible at the spatial extent of a river basin. Thus we suggest that following research focuses on
196
more river basins to try and assess universal drivers of movement, and thus potentially identify
key features for conservation.
Acknowledgments
We warmly thank Dr Van Strien for his help with the implementation of MLPE models in R.
Funding
This work was funded by the French Agency for Biodiversity (n° 136406), Electricité De France
(n°8619-5920011137).
References
Albouy, C., Guilhaumon, F., Villéger, S., Mouchet, M., Mercier, L., Culioli, J., Tomasini, J., Le Loc‘h, F.,
Mouillot, D., 2011. Predicting trophic guild and diet overlap from functional traits: statistics,
opportunities and limitations for marine ecology. Mar. Ecol. Prog. Ser. 436, 17–28.
https://doi.org/10.3354/meps09240
Altermatt, F., Fronhofer, E.A., 2018. Dispersal in dendritic networks: Ecological consequences on the
spatial distribution of population densities. Freshw. Biol. 63, 22–32.
https://doi.org/10.1111/fwb.12951
Baines, C.B., McCauley, S.J., Rowe, L., 2015. Dispersal depends on body condition and predation risk in
the semi-aquatic insect, Notonecta undulata. Ecol. Evol. 5, 2307–2316.
https://doi.org/10.1002/ece3.1508
Berry, O., Tocher, M.D., Sarre, S.D., 2004. Can assignment tests measure dispersal? Mol. Ecol. 13, 551–
561. https://doi.org/10.1046/j.1365-294X.2004.2081.x
Blair, C., Jiménez Arcos, V.H., Mendez de la Cruz, F.R., Murphy, R.W., 2013. Landscape genetics of Leaf-
Toed geckos in the tropical dry forest of northern Mexico. PLoS ONE 8, e57433.
https://doi.org/10.1371/journal.pone.0057433
Bonte, D., Van Dyck, H., Bullock, J.M., Coulon, A., Delgado, M., Gibbs, M., Lehouck, V., Matthysen, E.,
Mustin, K., Saastamoinen, M., Schtickzelle, N., Stevens, V.M., Vandewoestijne, S., Baguette, M.,
Barton, K., Benton, T.G., Chaput-Bardy, A., Clobert, J., Dytham, C., Hovestadt, T., Meier, C.M.,
Palmer, S.C.F., Turlure, C., Travis, J.M.J., 2012. Costs of dispersal. Biol. Rev. 87, 290–312.
https://doi.org/10.1111/j.1469-185X.2011.00201.x
Bowler, D.E., Benton, T.G., 2005. Causes and consequences of animal dispersal strategies: relating
individual behaviour to spatial dynamics. Biol. Rev. Camb. Philos. Soc. 80, 205–225.
Bracciali, C., Guzzo, G., Giacoma, C., Dean, J.M., Sarà, G., 2016. Fish functional traits are affected by
hydrodynamics at small spatial scale. Mar. Environ. Res. 113, 116–123.
https://doi.org/10.1016/j.marenvres.2015.12.002
Branch, G.M., 1975. Mechanisms reducing intraspecific competition in Patella Spp.: Migration,
differentiation and territorial behaviour. J. Anim. Ecol. 44, 575. https://doi.org/10.2307/3612
Bremset, G., Berg, O.K., 1997. Density, size-at-age, and distribution of young Atlantic salmon (Salmo
salar) and brown trout (Salmo trutta) in deep river pools. Can. J. Fish. Aquat. Sci. 54, 2827–2836.
https://doi.org/10.1139/f97-185
197
Brooker, R.W., Travis, J.M.J., Clark, E.J., Dytham, C., 2007. Modelling species‘ range shifts in a changing
climate: The impacts of biotic interactions, dispersal distance and the rate of climate change. J.
Theor. Biol. 245, 59–65. https://doi.org/10.1016/j.jtbi.2006.09.033
Brooks, A.J., Wolfenden, B., Downes, B.J., Lancaster, J., 2018. Barriers to dispersal: The effect of a weir
on stream insect drift. River Res. Appl. 34, 1244–1253. https://doi.org/10.1002/rra.3377
Cain, M.L., Milligan, B.G., Strand, A.E., 2000. Long-distance seed dispersal in plant populations. Am. J.
Bot. 87, 1217–1227. https://doi.org/10.2307/2656714
Caplat, P., Edelaar, P., Dudaniec, R.Y., Green, A.J., Okamura, B., Cote, J., Ekroos, J., Jonsson, P.R.,
Löndahl, J., Tesson, S.V., Petit, E.J., 2016. Looking beyond the mountain: dispersal barriers in a
changing world. Front. Ecol. Environ. 14, 261–268. https://doi.org/10.1002/fee.1280
Carrara, F., Rinaldo, A., Giometto, A., Altermatt, F., 2014. Complex interaction of dendritic connectivity
and hierarchical patch size on biodiversity in river-like landscapes. Am. Nat. 183, 13–25.
https://doi.org/10.1086/674009
Cayuela, H., Rougemont, Q., Prunier, J.G., Moore, J.-S., Clobert, J., Besnard, A., Bernatchez, L., 2018.
Demographic and genetic approaches to study dispersal in wild animal populations: A
methodological review. Mol. Ecol. 27, 3976–4010. https://doi.org/10.1111/mec.14848
Clarke, R.T., Rothery, P., Raybould, A.F., 2002. Confidence limits for regression relationships between
distance matrices: Estimating gene flow with distance. J. Agric. Biol. Environ. Stat. 7, 361–372.
https://doi.org/10.1198/108571102320
Clegg, S.M., Kelly, J.F., Kimura, M., Smith, T.B., 2003. Combining genetic markers and stable isotopes to
reveal population connectivity and migration patterns in a Neotropical migrant, Wilson‘s warbler
(Wilsonia pusilla). Mol. Ecol. 12, 819–830. https://doi.org/10.1046/j.1365-294X.2003.01757.x
Clobert, J. (Ed.), 2012. Dispersal ecology and evolution, 1st ed. ed. Oxford University Press, Oxford.
Clobert, J. (Ed.), 2001. Dispersal. Oxford University Press, Oxford ; New York.
Clobert, J., Le Galliard, J.-F., Cote, J., Meylan, S., Massot, M., 2009. Informed dispersal, heterogeneity in
animal dispersal syndromes and the dynamics of spatially structured populations. Ecol. Lett. 12,
197–209. https://doi.org/10.1111/j.1461-0248.2008.01267.x
Cockburn, A., Scott, M.P., Scotts, D.J., 1985. Inbreeding avoidance and male-biased natal dispersal in
Antechinus spp. (Marsupialia: Dasyuridae). Anim. Behav. 33, 908–915.
https://doi.org/10.1016/S0003-3472(85)80025-7
Cook, B.D., Bunn, S.E., Hughes, J.M., 2007. Molecular genetic and stable isotope signatures reveal
complementary patterns of population connectivity in the regionally vulnerable southern pygmy
perch (Nannoperca australis). Biol. Conserv. 138, 60–72.
https://doi.org/10.1016/j.biocon.2007.04.002
Cornuet, J.M., Piry, S., Luikart, G., Estoup, A., Solignac, M., 1999. New methods employing multilocus
genotypes to select or exclude populations as origins of individuals. Genetics 153, 1989–2000.
Cote, J., Clobert, J., 2007. Social personalities influence natal dispersal in a lizard. Proc. R. Soc. B Biol.
Sci. 274, 383–390. https://doi.org/10.1098/rspb.2006.3734
Cote, J., Clobert, J., Brodin, T., Fogarty, S., Sih, A., 2010. Personality-dependent dispersal:
characterization, ontogeny and consequences for spatially structured populations. Philos. Trans. R.
Soc. B Biol. Sci. 365, 4065–4076. https://doi.org/10.1098/rstb.2010.0176
Cote, Julien, Fogarty, S., Weinersmith, K., Brodin, T., Sih, A., 2010. Personality traits and dispersal
tendency in the invasive mosquitofish (Gambusia affinis). Proc. R. Soc. B Biol. Sci. 277, 1571–
1579. https://doi.org/10.1098/rspb.2009.2128
Cuinat, R., Heland, M., 1979. Observations sur la dévalaison d‘alevins de truite commune (Salmo trutta L.)
dans le Lissuraga. Bull. Fr. Piscic. 1–17. https://doi.org/10.1051/kmae:1979005
Dingemanse, N.J., Both, C., van Noordwijk, A.J., Rutten, A.L., Drent, P.J., 2003. Natal dispersal and
personalities in great tits (Parus major). Proc. R. Soc. B Biol. Sci. 270, 741–747.
https://doi.org/10.1098/rspb.2002.2300
198
Edelaar, P., Siepielski, A.M., Clobert, J., 2008. Matching habitat choice causes directed gene flow: a
neglected dimension in evolution and ecology. Evolution 62, 2462–2472.
https://doi.org/10.1111/j.1558-5646.2008.00459.x
Elliott, J.M., 1994. Quantitative ecology and the brown trout, Oxford series in ecology and evolution.
Oxford University Press, Oxford ; New York.
Ferrari, S., Cribari-Neto, F., 2004. Beta regression for modelling rates and proportions. J. Appl. Stat. 31,
799–815. https://doi.org/10.1080/0266476042000214501
Freeman, L.C., 1977. A set of measures of centrality based on betweenness. Sociometry 40, 35.
https://doi.org/10.2307/3033543
Froese, R., 2006. Cube law, condition factor and weight-length relationships: history, meta-analysis and
recommendations. J. Appl. Ichthyol. 22, 241–253. https://doi.org/10.1111/j.1439-
0426.2006.00805.x
Gouraud, V., Capra, H., Sabaton, C., Tissot, L., Lim, P., Vandewalle, F., Fahrner, G., Souchon, Y., 2008.
Long-term simulations of the dynamics of trout populations on river reaches bypassed by
hydroelectric installations-analysis of the impact of different hydrological scenarios. River Res.
Appl. 24, 1185–1205. https://doi.org/10.1002/rra.1129
Haddad, N.M., Brudvig, L.A., Clobert, J., Davies, K.F., Gonzalez, A., Holt, R.D., Lovejoy, T.E., Sexton,
J.O., Austin, M.P., Collins, C.D., Cook, W.M., Damschen, E.I., Ewers, R.M., Foster, B.L., Jenkins,
C.N., King, A.J., Laurance, W.F., Levey, D.J., Margules, C.R., Melbourne, B.A., Nicholls, A.O.,
Orrock, J.L., Song, D.-X., Townshend, J.R., 2015. Habitat fragmentation and its lasting impact on
Earth‘s ecosystems. Sci. Adv. 1, e1500052. https://doi.org/10.1126/sciadv.1500052
Hanski, I., 1998. Metapopulation dynamics. Nature 396, 41–49. https://doi.org/10.1038/23876
Hawkes, C., 2009. Linking movement behaviour, dispersal and population processes: is individual variation
a key? Movement behaviour, dispersal and population. J. Anim. Ecol. 78, 894–906.
https://doi.org/10.1111/j.1365-2656.2009.01534.x
Heland, M., 1980. La dévalaison des alevins de truite commune, Salmo trutta L. II. Activité des alevins
―dévalants‖ comparés aux sédentaires. Ann. Limnol. 16, 247–254.
https://doi.org/10.1051/limn/1980017
Hitt, N.P., Angermeier, P.L., 2008. Evidence for fish dispersal from spatial analysis of stream network
topology. J. North Am. Benthol. Soc. 27, 304–320. https://doi.org/10.1899/07-096.1
Hobson, K.A., 1999. Tracing origins and migration of wildlife using stable isotopes: a review. Oecologia
120, 314–326. https://doi.org/10.1007/s004420050865
Höjesjö, J., Økland, F., Sundström, L.F., Pettersson, J., Johnsson, J.I., 2007. Movement and home range in
relation to dominance; a telemetry study on brown trout Salmo trutta. J. Fish Biol. 70, 257–268.
https://doi.org/10.1111/j.1095-8649.2006.01299.x
Holt, R.D., Barfield, M., 2015. The influence of imperfect matching habitat choice on evolution in source–
sink environments. Evol. Ecol. 29, 887–904. https://doi.org/10.1007/s10682-015-9789-0
Hoset, K.S., Ferchaud, A.-L., Dufour, F., Mersch, D., Cote, J., Le Galliard, J.-F., 2011. Natal dispersal
correlates with behavioral traits that are not consistent across early life stages. Behav. Ecol. 22,
176–183. https://doi.org/10.1093/beheco/arq188
Ims, R.A., 1995. Movement patterns related to spatial structures, in: Hansson, L., Fahrig, L., Merriam, G.
(Eds.), Mosaic landscapes and ecological processes. Springer Netherlands, Dordrecht, pp. 85–109.
https://doi.org/10.1007/978-94-011-0717-4_4
Jacob, S., Bestion, E., Legrand, D., Clobert, J., Cote, J., 2015. Habitat matching and spatial heterogeneity of
phenotypes: implications for metapopulation and metacommunity functioning. Evol. Ecol. 29, 851–
871. https://doi.org/10.1007/s10682-015-9776-5
Johnson, M.L., Gaines, M.S., 1990. Evolution of Dispersal: Theoretical Models and Empirical Tests Using
Birds and Mammals. Annu. Rev. Ecol. Syst. 21, 449–480.
https://doi.org/10.1146/annurev.es.21.110190.002313
199
Jonsson, B., Jonsson, N., Hansen, L.P., 1991. Differences in life history and migratory behaviour between
wild and hatchery-reared Atlantic salmon in nature. Aquaculture 98, 69–78.
https://doi.org/10.1016/0044-8486(91)90372-E
Kot, M., Lewis, M.A., van den Driessche, P., 1996. Dispersal data and the spread of invading organisms.
Ecology 77, 2027–2042. https://doi.org/10.2307/2265698
Labonne, J., Ravigné, V., Parisi, B., Gaucherel, C., 2008. Linking dendritic network structures to
population demogenetics: The downside of connectivity. Oikos 117, 1479–1490.
https://doi.org/10.1111/j.0030-1299.2008.16976.x
Lin, Y.-T.K., Batzli, G.O., 2001. The influence of habitat quality on dispersal, demography, and population
dynamics of voles. Ecol. Monogr. 71, 245–275. https://doi.org/10.1890/0012-
9615(2001)071[0245:TIOHQO]2.0.CO;2
Lurz, P.W.W., Garson, P.J., Wauters, L.A., 1997. Effects of temporal and spatial variation in habitat quality
on red squirrel dispersal behaviour. Anim. Behav. 54, 427–435.
https://doi.org/10.1006/anbe.1996.0486
Maisse, G., Bagliniere, J.L., 1990. The biology of brown trout, Salmo trutta L., in the River Scorff,
Brittany: a synthesis of studies from 1973 to 1984. Aquac. Res. 21, 95–106.
https://doi.org/10.1111/j.1365-2109.1990.tb00385.x
Manel, S., Gaggiotti, O., Waples, R., 2005. Assignment methods: matching biological questions with
appropriate techniques. Trends Ecol. Evol. 20, 136–142. https://doi.org/10.1016/j.tree.2004.12.004
Martin, J., Bareille, G., Berail, S., Pecheyran, C., Daverat, F., Bru, N., Tabouret, H., Donard, O., 2013.
Spatial and temporal variations in otolith chemistry and relationships with water chemistry: a useful
tool to distinguish Atlantic salmon Salmo salar parr from different natal streams: Otolith chemistry
of Salmo salar. J. Fish Biol. 82, 1556–1581. https://doi.org/10.1111/jfb.12089
Matthysen, E., 2012. Multicausality of dispersal: a review, in: Clobert, J., Baguette, M., Benton, T.G.,
Bullock, J.M. (Eds.), Dispersal ecology and evolution. Oxford University Press, pp. 3–18.
https://doi.org/10.1093/acprof:oso/9780199608898.003.0001
McPeek, M.A., Holt, R.D., 1992. The evolution of dispersal in spatially and temporally varying
environments. Am. Nat. 140, 1010–1027. https://doi.org/10.1086/285453
Monard, A.-M., Duncan, P., 1996. Consequences of natal dispersal in female horses. Anim. Behav. 52,
565–579. https://doi.org/10.1006/anbe.1996.0198
Morin, P.A., Martien, K.K., Taylor, B.L., 2009. Assessing statistical power of SNPs for population
structure and conservation studies. Mol. Ecol. Resour. 9, 66–73. https://doi.org/10.1111/j.1755-
0998.2008.02392.x
Myers, J.H., Krebs, C.J., 1971. Genetic, behavioral, and reproductive attributes of dispersing field voles
Microtus pennsylvanicus and Microtus ochrogaster. Ecol. Monogr. 41, 53–78.
https://doi.org/10.2307/1942435
Nielsen, E.E., Hemmer-Hansen, J., Larsen, P.F., Bekkevold, D., 2009. Population genomics of marine
fishes: identifying adaptive variation in space and time. Mol. Ecol. 18, 3128–3150.
https://doi.org/10.1111/j.1365-294X.2009.04272.x
Niitepõld, K., Smith, A.D., Osborne, J.L., Reynolds, D.R., Carreck, N.L., Martin, A.P., Marden, J.H.,
Ovaskainen, O., Hanski, I., 2009. Flight metabolic rate and Pgi genotype influence butterfly
dispersal rate in the field. Ecology 90, 2223–2232.
O‘Riain, M.J., Jarvis, J.U.M., Faulkes, C.G., 1996. A dispersive morph in the naked mole-rat. Nature 380,
619–621. https://doi.org/10.1038/380619a0
Ovidio, M., Baras, E., Goffaux, D., Birtles, C., Philippart, J.C., 1998. Environmental unpredictability rules
the autumn migration of brown trout (Salmo trutta L.) in the Belgian Ardennes, in: Lagardère, J.-P.,
Anras, M.-L.B., Claireaux, G. (Eds.), Advances in invertebrates and fish telemetry. Springer
Netherlands, Dordrecht, pp. 263–274. https://doi.org/10.1007/978-94-011-5090-3_30
Paetkau, D., Calvert, W., Stirling, I., Strobeck, C., 1995. Microsatellite analysis of population structure in
Canadian polar bears. Mol. Ecol. 4, 347–354. https://doi.org/10.1111/j.1365-294X.1995.tb00227.x
200
Pakkasmaa, S., Piironen, J., 2001. Morphological differentiation among local trout (Salmo trutta)
populations. Biol. J. Linn. Soc. 72, 231–239. https://doi.org/10.1111/j.1095-8312.2001.tb01313.x
Patterson, T., Thomas, L., Wilcox, C., Ovaskainen, O., Matthiopoulos, J., 2008. State–space models of
individual animal movement. Trends Ecol. Evol. 23, 87–94.
https://doi.org/10.1016/j.tree.2007.10.009
Paz-Vinas, I., Blanchet, S., 2015. Dendritic connectivity shapes spatial patterns of genetic diversity: a
simulation-based study. J. Evol. Biol. 28, 986–994. https://doi.org/10.1111/jeb.12626
Paz-Vinas, I., Loot, G., Stevens, V.M., Blanchet, S., 2015. Evolutionary processes driving spatial patterns
of intraspecific genetic diversity in river ecosystems. Mol. Ecol. 24, 4586–4604.
https://doi.org/10.1111/mec.13345
Peterman, W.E., Connette, G.M., Semlitsch, R.D., Eggert, L.S., 2014. Ecological resistance surfaces predict
fine-scale genetic differentiation in a terrestrial woodland salamander. Mol. Ecol. 23, 2402–2413.
https://doi.org/10.1111/mec.12747
Peterson, E.E., Hoef, J.M.V., 2014. STARS : An ArcGIS toolset used to calculate the spatial information
needed to fit spatial statistical models to stream network data. J. Stat. Softw. 56.
https://doi.org/10.18637/jss.v056.i02
Peterson, E.E., Ver Hoef, J.M., Isaak, D.J., Falke, J.A., Fortin, M.-J., Jordan, C.E., McNyset, K.,
Monestiez, P., Ruesch, A.S., Sengupta, A., Som, N., Steel, E.A., Theobald, D.M., Torgersen, C.E.,
Wenger, S.J., 2013. Modelling dendritic ecological networks in space: an integrated network
perspective. Ecol. Lett. 16, 707–719. https://doi.org/10.1111/ele.12084
Pringle, C.M., 2001. Hydrologic connectivity and the management of biological reserves: a global
perspective. Ecol. Appl. 11, 981–998. https://doi.org/10.1890/1051-
0761(2001)011[0981:HCATMO]2.0.CO;2
Pritchard, J.K., Stephens, M., Donnelly, P., 2000. Inference of population structure using multilocus
genotype data. Genetics 155, 945–959.
Pusey, A.E., 1987. Sex-biased dispersal and inbreeding avoidance in birds and mammals. Trends Ecol.
Evol. 2, 295–299. https://doi.org/10.1016/0169-5347(87)90081-4
Rannala, B., Mountain, J.L., 1997. Detecting immigration by using multilocus genotypes. Proc. Natl. Acad.
Sci. 94, 9197–9201. https://doi.org/10.1073/pnas.94.17.9197
Ronce, O., 2007. How does it feel to be like a rolling stone? Ten questions about dispersal Evolution.
Annu. Rev. Ecol. Evol. Syst. 38, 231–253.
https://doi.org/10.1146/annurev.ecolsys.38.091206.095611
Roussel, J.M., Bardonnet, A., 2002. Habitat de la truite commune (Salmo trutta L.) pendant la période
juvénile en ruisseau : préférences, mouvements, variations journalières et saisonnières. Bull. Fr.
Pêche Piscic. 435–454. https://doi.org/10.1051/kmae:2002044
Row, J.R., Knick, S.T., Oyler-McCance, S.J., Lougheed, S.C., Fedy, B.C., 2017. Developing approaches
for linear mixed modeling in landscape genetics through landscape-directed dispersal simulations.
Ecol. Evol. 7, 3751–3761. https://doi.org/10.1002/ece3.2825
Saint-Pé, K., Blanchet, S., Tissot, L., Poulet, N., Plasseraud, O., Loot, G., Veyssière, C., Prunier, J.G.,
2018. Genetic admixture between captive-bred and wild individuals affects patterns of dispersal in
a brown trout (Salmo trutta) population. Conserv. Genet. 19, 1269–1279.
https://doi.org/10.1007/s10592-018-1095-2
Shafer, A.B.A., Poissant, J., Cote, S.D., Coltman, D.W., 2011. Does reduced heterozygosity influence
dispersal? A test using spatially structured populations in an alpine ungulate. Biol. Lett. 7, 433–
435. https://doi.org/10.1098/rsbl.2010.1119
Shreve, R.L., 1966. Statistical law of stream numbers. J. Geol. 74, 17–37. https://doi.org/10.1086/627137
Simmons, A.D., Thomas, C.D., 2004. Changes in dispersal during species‘ range expansions. Am. Nat.
164, 378–395. https://doi.org/10.1086/423430
201
Smith, S.M., Maughan, P.J., 2015. SNP Genotyping Using KASPar assays, in: Batley, J. (Ed.), Plant
Genotyping. Springer New York, New York, NY, pp. 243–256. https://doi.org/10.1007/978-1-
4939-1966-6_18
Sondermann, M., Gies, M., Hering, D., Schröder, M., Feld, C.K., 2015. Modelling the effect of in-stream
and terrestrial barriers on the dispersal of aquatic insect species: a case study from a Central
European mountain catchment. Fundam. Appl. Limnol. Arch. Für Hydrobiol. 186, 99–115.
https://doi.org/10.1127/fal/2015/0597
Souza, F.L., Raizer, J., da Costa, H.T.M., Martins, F.I., 2008. Dispersal of Phrynops geoffroanus
(Chelidae) in an urban river in Central Brazil. Chelonian Conserv. Biol. 7, 257–261.
https://doi.org/10.2744/CCB-0698.1
Szulkin, M., Sheldon, B.C., 2008. Dispersal as a means of inbreeding avoidance in a wild bird population.
Proc. R. Soc. B Biol. Sci. 275, 703–711. https://doi.org/10.1098/rspb.2007.0989
Tonkin, J.D., Altermatt, F., Finn, D.S., Heino, J., Olden, J.D., Pauls, S.U., Lytle, David.A., 2018. The role
of dispersal in river network metacommunities: Patterns, processes, and pathways. Freshw. Biol.
63, 141–163. https://doi.org/10.1111/fwb.13037
Travis, J.M.J., Delgado, M., Bocedi, G., Baguette, M., Bartoń, K., Bonte, D., Boulangeat, I., Hodgson, J.A.,
Kubisch, A., Penteriani, V., Saastamoinen, M., Stevens, V.M., Bullock, J.M., 2013. Dispersal and
species‘ responses to climate change. Oikos 122, 1532–1540. https://doi.org/10.1111/j.1600-
0706.2013.00399.x
Van Strien, M.J., Keller, D., Holderegger, R., 2012. A new analytical approach to landscape genetic
modelling: least-cost transect analysis and linear mixed models: new analytical approach in
landscape genetics. Mol. Ecol. 21, 4010–4023. https://doi.org/10.1111/j.1365-294X.2012.05687.x
Wang, I.J., Bradburd, G.S., 2014. Isolation by environment. Mol. Ecol. 23, 5649–5662.
https://doi.org/10.1111/mec.12938
Waples, R.S., Gaggiotti, O., 2006. What is a population? An empirical evaluation of some genetic methods
for identifying the number of gene pools and their degree of connectivity: WHAT IS A
POPULATION? Mol. Ecol. 15, 1419–1439. https://doi.org/10.1111/j.1365-294X.2006.02890.x
202
Supplementary material for Chapter V.
Appendix S5.1Ratios based on Bracciali et al, 2005; Albouy et al., 2011)
Trait number Measure
1 Head length
2 Body length
3 Caudal fin length
4 Eye diameter
5 Distance between snout and head-body junction
6 Nose tip height
7 Eye position
8 Head depth
9 Body height
10 Peduncle height
Ratios name Calculation
Pointiness of head 1/8
Body height 9/(1+2)
Back width 10/9
SurfProp 10*3/(1+2+3)
203
Appendix S5.2 PCA on morphometric ratios
204
Appendix S5.3. PCAs performed on (A) environmental predictors and on (B) habitat availability
predictors
A
B
205
Appendix S5.4: maps of a random selection of 10 movements detected for each river basin
Cartographic illustration of the possible spatial extent of these movements in each river basin
based on the random selection of 10 movements. Although this is far from representing the actual
number of movements detected, it enables to represent possible events, showing the possible
spatial extent of these movements.
206
Appendix S5.5: summary of the complete model explaining individual movement propensity (A)
and correlation matrix of response variables tested in this model (B)
A Movement propensity Estimate Std.Error z-value Pr(>|z|)
(Intercept) -1.65313 0.14257 -11.595 <2.00E-16 *** BVONE -0.19713 0.34052 -0.579 0.562659
BVSEUGE 0.70176 0.20943 3.351 0.000806 ***
scale(Taille) 0.08202 0.20969 0.391 0.695698 scale(MoprphAx1) 0.09561 0.19074 0.501 0.6162 scale(MoprphAx2) -0.02341 0.17115 -0.137 0.891225 scale(FultonK) 0.28357 0.25715 1.103 0.270139 scale(Heterozygosity) 0.09338 0.10628 0.879 0.379644 BVONE:scale(Taille) 0.53737 0.34363 1.564 0.117864 BVSEUGE:scale(Taille) 0.48108 0.28383 1.695 0.090081 .
BVONE:scale(MoprphAx1) 0.02449 0.34044 0.072 0.942654 BVSEUGE:scale(MoprphAx1) 0.13095 0.27849 0.47 0.638202 BVONE:scale(MoprphAx2) 0.54855 0.30765 1.783 0.074582 .
BVSEUGE:scale(MoprphAx2) 0.16483 0.21858 0.754 0.450795 BVONE:scale(FultonK) -3.74559 1.91545 -1.955 0.050528 .
BVSEUGE:scale(FultonK) 1.17945 1.27733 0.923 0.355817 BVONE:scale(Heterozygosity) 0.57336 0.41332 1.387 0.165383 BVSEUGE:scale(Heterozygosity) 0.11251 0.21045 0.535 0.592897
B Body length MoprphAx1 MoprphAx2 FultonK Heterozygosity
Body length 1 -0.63 -0.35 -0.12 -0.01
MoprphAx1 -0.63 1 -0.01 -0.04 -0.02
MoprphAx2 -0.35 -0.01 1 0.04 0.15
FultonK -0.12 -0.04 0.04 1 0
Heterozygosity -0.01 -0.02 0.15 0 1
207
Appendix S5.6: summary of the complete model explaining individual movement distance (A) and
correlation matrix of response variables tested in this model (B)
A Movement distance Estimate Std. Error t value Pr(>|t|)
(Intercept) 9.301329 0.09854 94.395 < 2e-16 ***
scale(Taille) -0.20453 0.14133 -1.447 0.1495 BVONE -1.79321 0.20075 -8.932 3.70E-16 ***
BVSEUGE -0.88851 0.12332 -7.205 1.34E-11 ***
scale(MoprphAx1) -0.18276 0.11349 -1.61 0.109 scale(MoprphAx2) 0.08745 0.11064 0.79 0.4303 scale(FultonK) 0.000995 0.06669 0.015 0.9881 scale(Heterozygosity) 0.067265 0.06833 0.984 0.3262 scale(Taille):BVONE 0.414534 0.20071 2.065 0.0403 *
scale(Taille):BVSEUGE -0.05933 0.17624 -0.337 0.7368 BVONE:scale(MoprphAx1) 0.141016 0.20501 0.688 0.4924 BVSEUGE:scale(MoprphAx1) 0.139922 0.14973 0.934 0.3512 BVONE:scale(MoprphAx2) -0.17694 0.1616 -1.095 0.2749 BVSEUGE:scale(MoprphAx2) -0.24829 0.14197 -1.749 0.0819 .
BVONE:scale(FultonK) 0.986305 0.6201 1.591 0.1134 BVSEUGE:scale(FultonK) -0.03389 0.11054 -0.307 0.7595 BVONE:scale(Heterozygosity) -0.31747 0.29357 -1.081 0.2809 BVSEUGE:scale(Heterozygosity) -0.01722 0.1157 -0.149 0.8818
B Body length MoprphAx1 MoprphAx2 FultonK Heterozygosity
Body length 1 -0.63 -0.35 -0.12 -0.01
MoprphAx1 -0.63 1 -0.01 -0.04 -0.02
MoprphAx2 -0.35 -0.01 1 0.04 0.15
FultonK -0.12 -0.04 0.04 1 0
Heterozygosity -0.01 -0.02 0.15 0 1
208
Appendix S5.7: summary of the complete model explaining pairwise movements between sites (A)
and correlation matrix of response variables tested in this model (B)
A Pairwise movements between sites Estimate Std. Error t value
(Intercept) 0.62268 0.2606 2.389
BVONE -0.62383 0.85032 -0.734
BVSEUGE 0.45996 0.36586 1.257
scale(PairwiseObstacles) 0.13235 0.36061 0.367
scale(PairwiseFlowConnect) 0.24562 0.26781 0.917
scale(PairwiseDist) 0.0848 0.25054 0.338
scale(dTempAx1) -0.11097 0.21026 -0.528
scale(dAltAx2) -0.07212 0.249 -0.29
scale(dStrahler) -0.11376 0.2392 -0.476
BVONE:scale(PairwiseObstacles) -0.19079 0.90387 -0.211
BVSEUGE:scale(PairwiseObstacles) -0.38614 0.50362 -0.767
BVONE:scale(PairwiseFlowConnect) 1.03918 0.46154 2.252
BVSEUGE:scale(PairwiseFlowConnect) 0.20898 0.37656 0.555
BVONE:scale(PairwiseDist) -0.40669 0.58698 -0.693
BVSEUGE:scale(PairwiseDist) -0.22123 0.5364 -0.412
BVONE:scale(dTempAx1) -0.43107 0.53606 -0.804
BVSEUGE:scale(dTempAx1) 0.13055 0.39322 0.332
BVONE:scale(dAltAx2) -0.74786 0.43178 -1.732
BVSEUGE:scale(dAltAx2) -0.54182 0.35789 -1.514
BVONE:scale(dStrahler) -0.66751 0.39548 -1.688
BVSEUGE:scale(dStrahler) -0.13522 0.32943 -0.41
B Obstacles FlowConnect Dist dTempAx1 dAltAx2 dStrahler
Obstacles 1 -0.21 0.72 0.41 0.31 0.03
FlowConnect -0.21 1 -0.37 -0.14 0.23 0.4
Dist 0.72 -0.37 1 0.37 0.34 -0.01
dTempAx1 0.41 -0.14 0.37 1 0 -0.05
dAltAx2 0.31 0.23 0.34 0 1 0.2
dStrahler 0.03 0.4 -0.01 -0.05 0.2 1
209
Appendix S5.8: summary of the complete model explaining proportion of immigrants (A) and
correlation matrix of response variables tested in this model (B)
A Proportion of immigrants Estimate
Std. Error z value Pr(>|z|)
(Intercept) -0.8468 1.2852 -0.659 0.51 TempAx1 0.4938 0.5963 0.828 0.4077 BVONE -5.3202 2.1937 -2.425 0.0153 *
BVSEUGE 3.2601 1.7403 1.873 0.061 .
AltAx2 -0.6013 0.6972 -0.862 0.3884 HabAvailAx1 -0.5564 0.5903 -0.943 0.3459 Strahler 0.3492 0.3684 0.948 0.3432 Centrality -0.6394 0.6463 -0.989 0.3225 TempAx1:BVONE 0.5943 0.9638 0.617 0.5375 TempAx1:BVSEUGE -0.8801 0.6624 -1.329 0.184 BVONE:AltAx2 2.0907 1.1604 1.802 0.0716 .
BVSEUGE:AltAx2 0.3404 0.7432 0.458 0.6469 BVONE:HabAvailAx1 2.0268 0.9629 2.105 0.0353 *
BVSEUGE:HabAvailAx1 1.8996 0.8278 2.295 0.0217 *
BVONE:Strahler 1.7005 0.912 1.865 0.0622 .
BVSEUGE:Strahler -0.8406 0.5864 -1.434 0.1517 BVONE:Centrality 1.5232 0.8693 1.752 0.0797 .
BVSEUGE:Centrality -0.4647 0.8476 -0.548 0.5835
B TempAx1 AltAx2 HabAvailAx1 Strahler Centrality
TempAx1 1 0 -0.46 -0.19 -0.07 AltAx2 0 1 -0.29 -0.52 -0.29 HabAvailAx1 -0.46 -0.29 1 0.47 0.29 Strahler -0.19 -0.52 0.47 1 0.31 Centrality -0.07 -0.29 0.29 0.31 1
210
General discussion
In this thesis, I developed new methodological approaches to study movement at the scale of the
river basin, and brought new insights in movement patterns of Brown trout, a key species in
headwater ecosystems. First, I developed a genetic tool useful for population genetic studies in the
brown trout, as well as hybridization with captive-bred strains. This Single Nucleotide
Polymorphism array mapped on the brown trout linkage map is cost effective, and especially,
should be a useful and universal tool for many population genetics studies on this species. This
tool was used in all (but one) subsequent chapters of my thesis. This tool exhibited high
genotyping success and enabled to successfully identify patterns of isolation-by-distance, yielding
similar patterns for population differentiation and admixture assessments than those obtained with
a panel of microsatellites but with better ability to detect introgression and differentiation. Then,
based on recent methods, I developed a procedure combining different markers to study movement
with improved accuracy and power. Using new methods of machine learning to combine them is a
very promising approach, and at the scales of the rivers studied in this PhD, these novel statistical
methods appear as an optimal way of investigating movements with high accuracy and power. In
particular, isotopes appeared as very efficient markers to discriminate and assign individuals to
their population of origin, and these non-lethal markers thus appear highly promising, as this is
one of their first successful applications for detecting movements at these spatial extents. The
second aim of my PhD was to identify biotic and abiotic factors underlying movement. However,
as stocking was shown to modify spatial patterns of genetic diversity and differentiation, which are
the base of movement detection, and thus interfere with movement inference, I first described the
effects of captive breeding on spatial patterns of genetic diversity, and found that captive-bred
genotypes increased diversity and differentiation. Moreover, I found that expected natural patterns
could be reversed in the case of higher captive-bred genotypes occurring at the level of
populations, thus showing that human induced changes can affect natural patterns therefore with
great implications. Further, I assessed how different genetic background (i.e. wild vs. carrying
captive-bred ancestry) created differential dispersal patterns, and found that this variable influence
movement propensity, movement distances, and the direction of movements, with the two strains
being highly parapatric. Finally, I aimed at better understanding how individual, environmental
and landscape related factors influence movements in natural populations, thus taking into
211
consideration only individuals that were supposedly wild. We found that body size was a main and
universal driver of movement patterns, and that the position of the site within riverscapes, and
habitat availability influenced the number of immigrants, although this depended on the river
considered. Finally, landscape features also influenced movement; we found that sites that are
directly connected by the water flow, and those that are similar in terms of elevation and stream
order exchanged more migrants.
This work was in large part funded to gather biological information which is useful in terms of
management. In this aspect, we focused on two main conundrums posed by human threats:
stocking and habitat fragmentation. Therefore, the main aspects which will be discussed here
concern the conundrum posed by stocking in both studying movements and assessing its effect on
this crucial facet of life history, and on the use of individual based methods for inferring
movements along with the methodological considerations arising from this discipline when using
different types of markers.
FACTORS INFLUENCING MOVEMENT IN THE BROWN TROUT, AND
MANAGEMENT/CONSERVATION IMPLICATIONS
Factors influencing movement
In Chapters 4 and 5, we found that there were drivers of movements which seem to be
generalizable, and others which are more context-dependent. Understanding how these relations
function thus has strong implications in understanding dynamics of brown trout populations.
First for individual drivers, it appeared that body length was a main determinant of both movement
propensity and movement distance. Indeed, in Chapter 5, we found that longer individuals tended
to be more prone to movements, and this was observed for all river basins. This was rather
unexpected, as generally, smaller individuals, which are younger, were widely shown to undergo
dispersal because their territory is not yet established (Andreu and Barba, 2006; Gachot-Neveu et
al., 2009). Moreover, in Chapter 4, although it was non-significant, there was an observed
tendency for this more expected pattern. Therefore, as the life cycle of the Brown trout is highly
complex and movements are closely linked to age and developmental stage, we suggest that the
change in spatial scale (Chapter 4 concerned very high headwaters only whereas Chapter 5
212
concentrated on a wider scale, although also including the higher reaches) could translate a
differential use of habitat depending on these time-scales.
These different spatio-temporal scales could also explain why in terms of movement distances, the
One River in its globality revealed that larger fish disperse further than smaller ones, whereas
when considering only the Neste d‘Oueil (its tributary, Chapter 4), the inverse relation was
observed: in this latter river, but also in the Aude and the Seuge Rivers, smaller fish exhibited
longer movements than larger ones. This observation is more consistent with what was observed in
riverine fish (Skalski and Gilliam, 2000): being resident can increase survival because it implies
not spending the costs for movement (Johnsson et al., 1999), thus larger and more dominant fish
may choose to hold on to their origin location, forcing smaller ones to move more (Vøllestad et al.,
2012). However, this scenario could concern only a small window of life history. Indeed, other
studies have shown that conversely, individuals which are larger, and in better body condition,
tend to be more mobile, in part because they can afford the cost of it. These two diverging
scenarios thus appear to be very dependent on the moment considered and when it arises within
the life cycle and within the habitat used at that moment: when considering post natal dispersal for
instance, smaller individuals should disperse more from higher reaches of the river, whereas when
considering foraging movements, or migrations, we suggest that larger individuals should exhibit a
more pronounced movement behavior.
At the population level, it appeared that several variables drove the number of immigrants which a
location is likely to receive. This is of high interest because it gives an insight into the underlying
mechanisms of source-sink dynamics in these systems. Also closely linked to the time and spatial
scales considered, we suggest that several mechanisms among which some were identified in our
study can play a crucial role in determining why a location should receive more immigrants than
others. First, the positional features (as discussed in chapter 5) such as Strahler index, centrality,
but also habitat availability which is closely linked to the distance to the river mouth were shown
to strongly influence movement patterns. These findings are congruent with the fact that larger,
more downstream and connected demes have been shown to generally act as sink populations
(Paz-Vinas et al., 2015). However, in our study, it appeared that while these positional
characteristics strongly drove movement patterns, features concerning environmental conditions,
in particular those in relation with a temperature gradient (i.e. temperature, dissolved oxygen, and
213
conductivity) were not linked to immigration. This is relatively surprising, as we could expect that
some habitats are more favorable for immigrants due to these physic-chemical features (Schlosser,
1998). However, this can be explained by what has been discussed previously: the dynamics of the
population i.e dispersal of fry, seasonal migration etc… could be way more important in shaping
movement patterns than local environmental conditions and the subsequent individual preferences
in terms of local suitability. Similarly, when investigating the number of movements between
sites, we also found that these positional variables were the only ones which influenced
movements (i.e. similar sites in terms of altitudinal axis and Strahler index exchanged more
individuals).
A main perspective in this direction would be to have a robust estimate of emigration. Indeed, to
investigate source-sink dynamics, or simply whether some locations can be unfavorable thus
tending to be emigrated from, having an accurate estimate of emigration would have been of high
interest. Our sampling resolution did not correspond to measuring this parameter accurately, as
discussed in chapter 5, therefore we suggest that direct methods could be the only way to track
emigrating individuals. This would be highly interesting, as we believe that the factors which drive
an individual to leave a location are as important as those driving immigration in understanding
metapopulation dynamics, particularly in a species which exhibits such a complex life cycle. In
particular, it would enable to assess the effect of inbreeding on the decision to emigrate, and also
the effect of population density as these parameters were shown to be important triggers of
movement, in particular, during the fry and parr stages. Indeed, density dependent emigration (as
individuals flee over populated locations in which competition for space and resource high) has
been observed (Bagliniere et al., 1989; Cuinat and Heland, 1979; Einum et al., 2006), and kin
interactions and inbreeding were also shown to modulate movements (Carlsson et al., 2004, Sanz
et al. 2011), either by concentrating related individuals, or conversely by separating them (kin
avoidance, Kasuya, 2000; Perrin and Mazalov, 1999).
Conservation
In terms of conservation and management in general, these finding have, as discussed in the last
chapter, strong implications, in particular because they may help to identify units of interest, at
different scales, and particularly because they can be very dependent on the context, or on the
214
contrary generalizable. First at the individual level, size was shown to strongly drive movements,
and depending on the context and on the spatial scale considered. In terms of management, the
Brown trout is a species which, especially in France, is very widely fished for recreation, and
deciding on the size at which an individual can be harvested is the topic of many arguments
between managers. Our results could suggest that taking movements dynamics into account could
perhaps add pertinence to these measures, for instance, protecting individuals which move most
could be of interest in preserving gene flow. In this case this would potentially imply that contrary
to what is usually done, protecting the large individuals from the downstream reaches could be
more important than protecting the smaller ones in those areas, whereas protection the small
individuals could be far more important in the upper reaches.
Second, at the landscape scale, the fact that positional features appear as the main drivers of
movement within a riverscape, and that it is possible to identify the underlying mechanisms of
what makes a population a source or a sink can also be useful in management. For instance, when
putting effort on dam passages, these findings could help to better integrate movement dynamics
in the prioritization of areas to protect/restore. For instance, we suggest that main sink populations
are of high conservation interest because they concentrate diversity, which is crucial at the scale of
a species (Groom et al., 2006). Therefore, in the rivers studied, the context-dependence of the
relation between positional features and source-sink dynamics should be carefully taken into
account for these matters. For instance in the One River, headwaters seem to receive a higher
proportion of migrants than lower reaches, therefore, protecting sink populations in this basin
would consist in protecting upper reaches, whereas in the Seuge and in the Aude, the sink
populations were located on lower reaches. In this aspect, studying source populations and as
discussed previously zones from which individuals emigrate would also have been of high interest
because these locations were shown to be crucial in shaping sink demes: in river systems, these
lowly diverse and highly differentiated locations generate diversity in the downstream sink
populations (Finn et al., 2011) and are thus also highly important in the perspective of
conservation, thus assessing this in our study systems remains to be done.
215
STUDYING MOVEMENT IN A MANAGED AND STOCKED SPECIES
The brown trout is an emblematic species that has been stocked for decades, mainly for sustaining
recreational fisheries. Assessing the short and long term consequences of this practice is a
prerequisite for any study aiming at investigating ―natural‖ processes in riverscapes, because
stocking has profoundly affected this species. For instance in chapter 5, we first discarded
individuals of captive-bred ancestry to investigate natural patterns of movement, however, even if
in chapter 3, we found that there tended to be a strong spatial segregation between wild and
captive bred strains, with relatively low admixture rates, it was widely shown that they can interact
(Blanchet et al., 2008). Therefore, even by discarding individuals exhibiting captive-bred ancestry,
natural patterns in wild populations can still strongly be affected by the presence of these non-local
strains, in the same way that another species would. For instance, Blanchet et al. (2008) showed
that captive-bred individuals revealed higher aggressiveness and directly compete with wild
individuals. Moreover, as these strains were, in our case studies, mainly located in the upper
reaches of the basins, we could also suggest that, given the impact of Brown trout on food
resource, they could be affecting the lower reaches, in particular because of excretion and
consummation of resource which may have drifted downstream without their presence (Morita et
al., 2004; Moyle, 1986; Townsend, 1996). Among the effects that these exogenous strains may
potentially have on wild ones, a main one we observed in these works was the modulation of
dispersal patterns through their underlying mechanisms (Chapter 4). Therefore, studying
movement in the context of stocking showed that the underlying mechanisms of movement are
strongly affected by stocking.
On top of these biological effects on movements for wild populations, stocking is also a
methodological conundrum in the detection of movements using genetics. As discussed in chapter
3, we suggest that all studies using genetics are likely to be biased by the presence of these
captive-bred genotypes because massive displacements and introductions have most likely
strongly affected population structure, and as shown in chapter 3, patterns of genetic diversity and
differentiation. This therefore implies that for instance, studying movements based on genetics
only may be somewhat erroneous, although it appears that captive-bred and wild strains seem to
remain relatively isolated. This is also why using other types of data, and in particular isotopes
could be the powerful way to go around these biases. However, behavioral differences between
216
strains could also lead to different feeding strategies, which would reflect on isotopic signatures in
tissues (except in those which take the signature of water such as otoliths). The brown trout is a
very interesting model, but universally affected by humans, and thus drawing conclusions on
natural processes in this species is therefore very difficult. Studies must now integrate the fact that
populations have been widely modified, and developing tools which could overcome or at least
take into account these considerations is crucial.
In chapter one, the tool developed is suggested to function rather universally, and I demonstrated
that it was actually the case in two lineages of Brown trout. The study systems in this PhD consist
in different and independent river basins, thus it was crucial that all basins were studied using the
same approach and tools to draw both methodological and biological conclusions. Indeed,
replicability is often a standard for all scientific investigations, and we suggest that this tool will be
used to compare findings from other rivers to the ones we studied. Moreover, the
straightforwardness of these types of tools is a requisite for their use in conservation, particularly
because managers need tools that are not time consuming and which they can extract information
from simply. Indeed, this tool is highly advantageous in terms of efficiency to put in place: they
require only sampling tissue, encapsulating the sample in ethanol and sent samples to the
genotyping platform. Within a few weeks, and without having to perform DNA extraction because
the sampling platforms generally offer this step at a very reasonable cost, the data is sent back
ready to be analyzed. Additionally, each research group generally uses its own panel of
microsatellites for these types of studies, which makes comparisons among studies relatively
difficult. For instance, a global study gathering all generated datasets on the Brown trout would be
impossible for this reason. Therefore, the SNP panel developed here appears as a straightforward
tool which has the potential to now be used commonly for all next population genetics on this
species, which is why we call for its application all across the species‘ range.
217
INSIGHTS OF DETECTING MOVEMENTS WITH INDIVIDUAL BASED
METHODS
Assignment methods and combinations
In this thesis, the tools used to discriminate between populations i.e. genetics and isotopes,
revealed unexpected features at the spatial extent which we considered. First, genetics were not as
discriminant and thus accurate as we expected for this species which generally exhibits strong
genetic structure even at small spatial extents (Swatdipong et al., 2010; Vøllestad et al., 2012;
Berrebi, 2015). However, the basins considered were only studied in their headwaters, thus the
scale may be too small to detect this structure. Similarly, Cook et al., (2007) found similar lack of
genetic structure in pygmy perch at relatively similar scales, resulting in relatively low assignment
of individuals to their capture site, however, they found that stable isotopes of Carbon and
Nitrogen from fin tissue on the other hand were much more discriminant. This is particularly
congruent with our results, notably in the Aude River. Interestingly, we found very few other
studies which go in this direction as generally, isotopes are used for assignment purposes at much
wider and contrasted scales. However, there are some studies showing that isotopic signatures of
local dietary origin, as in our case nitrogen and carbon, can be informative over relatively small
spatial and temporal scales for some species (Cunjak et al., 2005; Maruyama et al., 2001) because
they can vary over small spatial scales (Peterson and Fry, 1987). But in terms of sufficient power
to detect movements, we expected that these markers would only provide high resolution at very
broad geographical extents, and even when combined (accordingly to Clegg et al., 2003; Gómez-
Díaz and González-Solis, 2007; Hall and Beissinger, 2017). This was not the case, as we found
that fine‐scale population resolution was satisfactory enough to make precise inferences about
movements, even though using our permutation process to make sure that assignment power is
sufficient, we lost some individuals. Another alternative to C and N could be trace elements as
they were shown to be even more sensitive to micro geographical differences and were already
shown to give even better assignment results at local scales, at least for birds, such as the example
in Gómez-Díaz and González-Solis, 2007; Hall and Beissinger (2017), where the authors assigned
Cory‘s Shearwaters caught in western Mediterranean long-liners to breeding colonies in Menorca,
Ibiza and Crete.
218
Comparison and significance of movement detection using different markers
As these markers displayed different discriminatory power, we suggest that they may be very
complementary tracers of animal movement especially because they concern different time scales,
and therefore have the potential to detect different types of movements (Clegg et al., 2003;
Rubenstein and Hobson, 2004). Additionally, the accuracy of detection of each type of movement
will also depend on the discriminatory power of a given marker for a given context (as seen in
figure 1). For instance, Cook 2007 found that classification to capture site was stronger for stable
isotope data than genetics in a case study river, suggesting that short-term, small-scale population
isolation is not maintained over longer timescales (Skalski and Gilliam, 2000).
Figure D1: Proportions of migrants, residents and unknown detected with genetics alone, isotopes alone,
and the combination of genetics and isotopes
219
Thus, the combined use of molecular markers and natural abundance isotopic signatures of
nitrogen and carbon has the potential to be a powerful approach for studying movements and
connectivity, and could enable to capture evens which would be missed by only considering one
data type. Indeed, some authors suggest that information contained in genetic and isotope markers
is largely additive and that therefore combining them is highly beneficial to increasing assignment
(Clegg et al., 2003; Webster et al., 2002). Combining genetics with isotopes is therefore used more
and more frequently. However, the fact that different types of movement can be detected
depending on the markers used, in our eyes, also poses a conundrum for their use in combination,
and very surprisingly, this consideration has to our knowledge never been discussed in these
studies: in two of the chapters (2 and 4), I used these combinations of different data types do detect
movements, and accordingly, I show that statistically, combining markers enabled a gain in
discrimination among populations, and therefore in accuracy and power for assignment. However,
as it is discussed in chapter 2, combining datasets may have some drawbacks, and we believe there
is a counterpart for this increase in power: the potential loss of biologically information of interest.
Indeed, the fact that each data sets enable to capture specific timescales (Hall and Beissinger,
2017) is, in the case of combination, not accounted for. Indeed, as discussed in chapter 2, the
genetic signature of an individual is permanent but isotopic signature can change over time
depending on diet and turnover rate (Hobson and Wassenaar, 2008). Therefore, assignment based
on isotopes will depend on the time elapsed between actual movement and time of capture, and
thus there are several possibilities of detecting individuals as migrants or residents according to the
marker used, with two possible mismatches due to these differences (Figure 2).
220
Figure D2: Biological and statistical significance of detecting individuals as residents with both genetics
and isotopes (bottom left), as residents according to genetics but migrants according to isotopes (top left),
as migrants according to genetics but residents according to isotopes (bottom right), and as migrants using
both markers (top right). There are two possible types of errors: Type I errors consist in assigning a
resident individual as a migrant; Type II errors consist in assigning a migrant as a resident.
Because these markers detect different movements, we therefore found that different individuals
may be assigned differently according to the marker considered (Figure 3). And this has very
strong implications on the legitimity of combining these datasets: in theory, an individual will be
assigned according to the marker which is most discriminant. Thus, in the Aude for instance, if an
individual is considered migrant according to isotopes, which for this basin are the most powerful,
but resident considering genetics, it is most likely that when combining the markers, the individual
will be considered as migrant because this decision is mainly driven by isotopes. And both of these
assignments may be highly interesting, but the combination does not allow investigating these
considerations: indeed, this could mean that the individual in question has returned to his capture
(and origin) site after having emigrated to a location where its isotopic signature changed.
Implications can thus vary considerably, because such an individual would be miss assigned in
consideration of its genetic origin.
221
Figure D3: Proportions of individuals detected as migrants with both genetics and isotpes (dark grey), of
individuals detected as residents with both genetics and isotopes (white), of individuals detected as
residents according to genetics but migrants according to isotopes (light gray), and of individuals detected
as migrants according to genetics but residents according to isotopes (intermediate gray)
In addition, other different scenarios emerge from these considerations, for instance, if an
individual moves from a location to another, during this transience, it will get closer to its
destination locations, thus isotopic signatures get closer while genetics do not change. This would
mean that when an individual arrived recently, the distance moved according to genetics would be
―true‖ whereas distance measured with isotopes should be shorter, because the individual has the
isotopic signature of an intermediate site (between departure and destination site). Therefore, if an
individual is detected as migrant according to genetics and also to isotopes, the site to which it is
assigned can differ, although both markers consider the individual as a migrant. In our results, we
222
found quite a few cases following this scenario, but not all cases. This thus suggests that a wide
variety of scenarios and mechanisms are probably taking place, and probably simultaneously.
Although they appear very intricate and question a large number of assumptions, among which
some developed in this thesis, these considerations open new insights as to studying different types
of movements, using different types of markers. Indeed, as detailed previously, the life cycle of the
Brown trout is extremely complex, and presents a continuum of different possibilities regarding
life-history strategies. Therefore, using markers such as genetics and isotopes, but also if possible
coupled to direct tracking of individuals could be the direction to take when studying such
complex species. Moreover, as we found that movement patterns are in some aspects very context
dependent, using various methods in a large number of study systems should be the most accurate
way to disentangle these different strategies, at different time and spatial scales.
CONCLUSION AND PERSPECTIVES
In this thesis, I answered both a methodological goal, by developing a universal and efficient
genetic tool, and assignment approaches to study movement in a machine learning framework.
Then, I used these tools to first assess the impact of stocking on patterns of genetic diversity and
differentiation, and, after assessing the effects of stocking on dispersal, I investigated the
individual, environmental and landscape linked drivers of movements in different rivers. These
works have also opened the door for many other questions about movement in the context of
anthropic perturbations on natural habitats, and due to financial and temporal limitations, many
aspects remain to be investigated. In particular, at these relatively large scales, it would be highly
interesting to almost exhaustively monitor all individuals from different independent catchments,
both in terms of biological findings, but also to enable a comparison of what it actually happening
and what we found using indirect methods. In this optic, the genetic data of Chapter IV in the
Neste d‘Oueil have been used in combination with PIT tagging (Alp et al. in prep). Similarly,
methods based on family reconstruction can also give very high definition information in
particular in the study of dispersal of juveniles. This has been done in chapter IV, but in other
basins, we suggest that the sampling design was too scarce to accurately use this approach, thus
sampling smaller scale basins more thoroughly could have also been an interesting approach.
223
An interesting perspective from the markers used is that besides studying movements, they could
enable to investigate many other questions. First, genetics are a tool which is accurate to
investigate the genetic ―health‖ of populations (without considering the effects of stocking),
through calculation of effective size, inbreeding (although it appeared very low in the populations
sampled). Furthermore, the genetic tool developed could also, by using different filters than those
aiming at relative neutrality of markers, be used to investigate patterns of selection for instance
using genome wide association studies. Second, with the data gathered, isotopes also still offer
many unexplored perspectives fur future studies. For instance, investigating different trophic
interactions within the river catchments of our study systems has not been done mainly because of
temporal limitations. In particular, we showed that stocking affected natural populations in many
ways (through genetic spatial patterns, and movement), thus investigating differences in diet, and
in trophic position between wild and captive-bred strains, and competition for food resource
between them could be another interesting way of assessing the effects of stocking on these
populations. Finally, morphometrics were globally found to be relatively disappointing in our
study systems, both for discriminating among populations and for explaining movement patterns at
the individual level. Therefore, we suggested that this tool was most likely lacking spatial
resolution. However, other studies have found it highly useful, thus perhaps using them differently
by calculating other ratios than those used, or measuring variables which we ignored, in particular
colorimetry or punctuation of the fish could have made these markers more exploitable.
Despite these considerations, our findings have brought insight of interest into the dynamics of the
species studied. Indeed, for subsequent studies on movement, the new tools developed have shown
to be efficient at the spatial scales of interest for management and conservation decisions, which
was lacking compared to other studies. Moreover, the biological findings in this thesis are also of
interest first because they open insights into the study of movement in a managed and stocked
species, thus more globally in the context of human induced perturbations on natural systems.
Finally, the identification of individual features, but also environments and landscape
characteristics which shape movement patterns has participated to knowledge in this already
widely studied species, which are also useful for management, and in particular for the definition
of interesting units for conservation prioritization.
224
References Aarestrup, K., Okland, F., Hansen, M.M., Righton, D., Gargan, P., Castonguay, M., Bernatchez, L.,
Howey, P., Sparholt, H., Pedersen, M.I., McKinley, R.S., 2009. Oceanic spawning migration of the
European Eel (Anguilla anguilla). Science 325, 1660–1660.
https://doi.org/10.1126/science.1178120
Acolas, M.-L., Roussel, J.-M., Baglinière, J.-L., 2008. Linking migratory patterns and diet to reproductive
traits in female brown trout (Salmo trutta L.) by means of stable isotope analysis on ova. Ecol.
Freshw. Fish 17, 382–393. https://doi.org/10.1111/j.1600-0633.2007.00290.x
Alanärä, A., Burns, M.D., Metcalfe, N.B., 2001. Intraspecific resource partitioning in brown trout: the
temporal distribution of foraging is determined by social rank. J. Anim. Ecol. 70, 980–986.
https://doi.org/10.1046/j.0021-8790.2001.00550.x
Alphandéry, P., Fortier, A., Sourdril, A., 2012. Les données entre normalisation et territoire : la
construction de la trame verte et bleue. Dév. Durable Territ.
https://doi.org/10.4000/developpementdurable.9282
Andreu, J., Barba, E., 2006. Breeding dispersal of Great Tits Parus major in a homogeneous habitat: effects
of sex, age, and mating status 14.
Aparicio, E., Rocaspana, R., de Sostoa, A., Palau-Ibars, A., Alcaraz, C., 2018. Movements and dispersal of
brown trout (Salmo trutta Linnaeus, 1758) in Mediterranean streams: influence of habitat and biotic
factors. PeerJ 6, e5730. https://doi.org/10.7717/peerj.5730
Araki, H., Schmid, C., 2010. Is hatchery stocking a help or harm? Aquaculture 308, S2–S11.
https://doi.org/10.1016/j.aquaculture.2010.05.036
Arlinghaus, R., Mehner, T., 2005. Determinants of management preferences of recreational anglers in
Germany: Habitat management versus fish stocking. Limnologica 35, 2–17.
https://doi.org/10.1016/j.limno.2004.10.001
Armstrong, J.D., Braithwaite, V.A., Huntingford, F.A., 1997. Spatial strategies of wild atlantic salmon parr:
exploration and settlement in unfamiliar areas. J. Anim. Ecol. 66, 203. https://doi.org/10.2307/6022
Bachman, R.A., 1984. Foraging behavior of free-ranging wild and hatchery brown trout in a stream. Trans.
Am. Fish. Soc. 113, 1–32. https://doi.org/10.1577/1548-8659(1984)113<1:FBOFWA>2.0.CO;2
Bacon, I.E., Hurly, T.A., Healy, S., 2010. Both the past and the present affect risk-sensitive decisions of
foraging rufous hummingbirds. Behav. Ecol. 21, 626–632. https://doi.org/10.1093/beheco/arq031
Baglinière, J.L., 1999. Introduction: The brown trout (Salmo trutta L)—its origin, distribution and
economic and scientific significance, in: Baglinière, J.L., Maisse, G. (Eds.), Biology and Ecology
of the Brown and Sea Trout. Springer London, London, pp. 1–12. https://doi.org/10.1007/978-1-
4471-0775-0_1
Bagliniere, J.L., Maisse, G., Lebail, P.Y., Nihouarn, A., 1989. Population dynamics of brown trout, Salmo
trutta L., in a tributary in Brittany (France): spawning and juveniles. J. Fish Biol. 34, 97–110.
https://doi.org/10.1111/j.1095-8649.1989.tb02960.x
Baguette, M., Blanchet, S., Legrand, D., Stevens, V.M., Turlure, C., 2013. Individual dispersal, landscape
connectivity and ecological networks: Dispersal, connectivity and networks. Biol. Rev. 88, 310–
326. https://doi.org/10.1111/brv.12000
Baguette, M., Van Dyck, H., 2007. Landscape connectivity and animal behavior: functional grain as a key
determinant for dispersal. Landsc. Ecol. 22, 1117–1129. https://doi.org/10.1007/s10980-007-9108-
4
Barton, K.A., Phillips, B.L., Morales, J.M., Travis, J.M.J., 2009. The evolution of an ‗intelligent‘ dispersal
strategy: biased, correlated random walks in patchy landscapes. Oikos 118, 309–319.
https://doi.org/10.1111/j.1600-0706.2008.16936.x
Barton, R.A., Whiten, A., Strum, S.C., Byrne, R.W., Simpson, A.J., 1992. Habitat use and resource
availability in baboons. Anim. Behav. 43, 831–844. https://doi.org/10.1016/S0003-3472(05)80206-
4
225
Bearhop, S., Furness, R.W., Hilton, G.M., Votier, S.C., Waldron, S., 2003. A forensic approach to
understanding diet and habitat use from stable isotope analysis of (avian) claw material. Funct.
Ecol. 17, 270–275. https://doi.org/10.1046/j.1365-2435.2003.00725.x
Begon, M., Harper, J.L., Townsend, C.R., 1988. Ecology: individuals, populations and communities, Repr.
with corr. 1988. ed. Blackwell, Oxford.
Beier, P., Noss, R.F., 1998. Do habitat corridors provide connectivity? Conserv. Biol. 12, 1241–1252.
https://doi.org/10.1111/j.1523-1739.1998.98036.x
Bekkevold, D., Helyar, S.J., Limborg, M.T., Nielsen, E.E., Hemmer-Hansen, J., Clausen, L.A.W.,
Carvalho, G.R., 2015. Gene-associated markers can assign origin in a weakly structured fish,
Atlantic herring. ICES J. Mar. Sci. J. Cons. 72, 1790–1801. https://doi.org/10.1093/icesjms/fsu247
Bembo, D.G., Beverton, R.J.H., Weightman, A.J., Cresswell, R.C., 1993. Distribution, growth and
movement of River Usk brown trout (Salmo trutta). J. Fish Biol. 43, 45–52.
https://doi.org/10.1111/j.1095-8649.1993.tb01178.x
Bernatchez, L., 2001. The evolutionary history of brown trout (Salmo trutta L.) inferred from
phylogeographic, nested clade, and mismatch analyses of mitochondrial dna variation. Evolution
55, 351. https://doi.org/10.1554/0014-3820(2001)055[0351:TEHOBT]2.0.CO;2
Berrebi, P., 2015. Three brown trout Salmo trutta lineages in Corsica described through allozyme variation:
salmo trutta genetics in corsica. J. Fish Biol. 86, 60–73. https://doi.org/10.1111/jfb.12534
Berrebi, P., Caputo Barucchi, V., Splendiani, A., Muracciole, S., Sabatini, A., Palmas, F., Tougard, C.,
Arculeo, M., Marić, S., 2019. Brown trout (Salmo trutta L.) high genetic diversity around the
Tyrrhenian Sea as revealed by nuclear and mitochondrial markers. Hydrobiologia 826, 209–231.
https://doi.org/10.1007/s10750-018-3734-5
Berrebi, P., Poteaux, C., Fissier, M., Cattaneo-Berrebi, G., 2000. Stocking impact and allozyme diversity in
brown trout from Mediterranean southern France. J. Fish Biol. 56, 949–960.
https://doi.org/10.1111/j.1095-8649.2000.tb00884.x
Berry, O., Tocher, M.D., Sarre, S.D., 2004. Can assignment tests measure dispersal? Mol. Ecol. 13, 551–
561. https://doi.org/10.1046/j.1365-294X.2004.2081.x
Berthold, P., Helbig, A.J., 2008. The genetics of bird migration: stimulus, timing, and direction. Ibis 134,
35–40. https://doi.org/10.1111/j.1474-919X.1992.tb04731.x
Blanchet, S., Páez, D.J., Bernatchez, L., Dodson, J.J., 2008. An integrated comparison of captive-bred and
wild Atlantic salmon (Salmo salar): Implications for supportive breeding programs. Biol. Conserv.
141, 1989–1999. https://doi.org/10.1016/j.biocon.2008.05.014
Blanchet, S., Rey, O., Etienne, R., Lek, S., Loot, G., 2010. Species-specific responses to landscape
fragmentation: implications for management strategies. Evol. Appl. 3, 291–304.
https://doi.org/10.1111/j.1752-4571.2009.00110.x
Blomqvist, D., Pauliny, A., Larsson, M., Flodin, L.-Å., 2010. Trapped in the extinction vortex? Strong
genetic effects in a declining vertebrate population. BMC Evol. Biol. 10, 33.
https://doi.org/10.1186/1471-2148-10-33
Bocherens, H., Drucker, D., 2003. Trophic level isotopic enrichment of carbon and nitrogen in bone
collagen: case studies from recent and ancient terrestrial ecosystems. Int. J. Osteoarchaeol. 13, 46–
53. https://doi.org/10.1002/oa.662
Bohling, J.H., 2016. Strategies to address the conservation threats posed by hybridization and genetic
introgression. Biol. Conserv. 203, 321–327. https://doi.org/10.1016/j.biocon.2016.10.011
Bonte, D., Van Dyck, H., Bullock, J.M., Coulon, A., Delgado, M., Gibbs, M., Lehouck, V., Matthysen, E.,
Mustin, K., Saastamoinen, M., Schtickzelle, N., Stevens, V.M., Vandewoestijne, S., Baguette, M.,
Barton, K., Benton, T.G., Chaput-Bardy, A., Clobert, J., Dytham, C., Hovestadt, T., Meier, C.M.,
Palmer, S.C.F., Turlure, C., Travis, J.M.J., 2012. Costs of dispersal. Biol. Rev. 87, 290–312.
https://doi.org/10.1111/j.1469-185X.2011.00201.x
Bowler, D.E., Benton, T.G., 2005. Causes and consequences of animal dispersal strategies: relating
individual behaviour to spatial dynamics. Biol. Rev. Camb. Philos. Soc. 80, 205–225.
226
Bradbury, I.R., Hamilton, L.C., Dempson, B., Robertson, M.J., Bourret, V., Bernatchez, L., Verspoor, E.,
2015. Transatlantic secondary contact in Atlantic Salmon, comparing microsatellites, a single
nucleotide polymorphism array and restriction-site associated DNA sequencing for the resolution
of complex spatial structure. Mol. Ecol. 24, 5130–5144. https://doi.org/10.1111/mec.13395
Branch, G.M., 1975. Mechanisms reducing intraspecific competition in Patella Spp.: migration,
differentiation and territorial behaviour. J. Anim. Ecol. 44, 575. https://doi.org/10.2307/3612
Breiman, L., 2001. [No title found]. Mach. Learn. 45, 5–32. https://doi.org/10.1023/A:1010933404324
Brodersen, J., Nilsson, P.A., Hansson, L.-A., Skov, C., Brönmark, C., 2008. Condition-dependent
individual decision-making determines cyprinid partial migration. Ecology 89, 1195–1200.
https://doi.org/10.1890/07-1318.1
Brönmark, C., Hulthén, K., Nilsson, P.A., Skov, C., Hansson, L.-A., Brodersen, J., Chapman, B.B., 2014.
There and back again: migration in freshwater fishes. Can. J. Zool. 92, 467–479.
https://doi.org/10.1139/cjz-2012-0277
Brönmark, C., Skov, C., Brodersen, J., Nilsson, P.A., Hansson, L.-A., 2008. Seasonal migration determined
by a trade-off between predator avoidance and growth. PLoS ONE 3, e1957.
https://doi.org/10.1371/journal.pone.0001957
Brown, J.H., Kodric-Brown, A., 1977. Turnover rates in insular biogeography: effect of immigration on
extinction. Ecology 58, 445–449. https://doi.org/10.2307/1935620
Burrell, K.H., Isely, J.J., Bunnell, D.B., Van Lear, D.H., Dolloff, C.A., 2000. Seasonal movement of brown
trout in a southern Appalachian river. Trans. Am. Fish. Soc. 129, 1373–1379.
https://doi.org/10.1577/1548-8659(2000)129<1373:SMOBTI>2.0.CO;2
Busst, G.M.A., Britton, J.R., 2018. Tissue-specific turnover rates of the nitrogen stable isotope as functions
of time and growth in a cyprinid fish. Hydrobiologia 805, 49–60. https://doi.org/10.1007/s10750-
017-3276-2
Butler, J.R.A., Radford, A., Riddington, G., Laughton, R., 2009. Evaluating an ecosystem service provided
by Atlantic salmon, sea trout and other fish species in the River Spey, Scotland: The economic
impact of recreational rod fisheries. Fish. Res. 96, 259–266.
https://doi.org/10.1016/j.fishres.2008.12.006
Byers, J.A., 2001. Correlated Random Walk equations of animal dispersal resolved by simulation. Ecology
82, 1680. https://doi.org/10.2307/2679810
Cabana, G., Rasmussen, J.B., 1996. Comparison of aquatic food chains using nitrogen isotopes. Proc. Natl.
Acad. Sci. 93, 10844–10847. https://doi.org/10.1073/pnas.93.20.10844
Campbell Grant, E.H., Lowe, W.H., Fagan, W.F., 2007. Living in the branches: population dynamics and
ecological processes in dendritic networks. Ecol. Lett. 10, 165–175. https://doi.org/10.1111/j.1461-
0248.2006.01007.x
Caraco, T., 1981. Risk-sensitivity and foraging groups. Ecology 62, 527–531.
https://doi.org/10.2307/1937716
Caraco, T., Blanckenhorn, W.U., Gregory, G.M., Newman, J.A., Recer, G.M., Zwicker, S.M., 1990. Risk-
sensitivity: ambient temperature affects foraging choice. Anim. Behav. 39, 338–345.
https://doi.org/10.1016/S0003-3472(05)80879-6
Carlsson, J., Carlsson, J.E.L., Olsén, K.H., Hansen, M.M., Eriksson, T., Nilsson, J., 2004. Kin-biased
distribution in brown trout: an effect of redd location or kin recognition? Heredity 92, 53–60.
https://doi.org/10.1038/sj.hdy.6800376
Carrara, F., Rinaldo, A., Giometto, A., Altermatt, F., 2014. Complex interaction of dendritic connectivity
and hierarchical patch size on biodiversity in river-like landscapes. Am. Nat. 183, 13–25.
https://doi.org/10.1086/674009
Castellón, T.D., Sieving, K.E., 2006. An experimental test of matrix permeability and corridor use by an
endemic understory bird: habitat connectivity translocation experiment. Conserv. Biol. 20, 135–
145. https://doi.org/10.1111/j.1523-1739.2006.00332.x
227
Caudron, A., Champigneulle, A., Guyomard, R., 2006. Assessment of restocking as a strategy for
rehabilitating a native population of brown trout Salmo trutta L. in a fast-flowing mountain stream
in the northern French Alps. J. Fish Biol. 69, 127–139. https://doi.org/10.1111/j.1095-
8649.2006.01156.x
Cayuela, H., Rougemont, Q., Prunier, J.G., Moore, J.-S., Clobert, J., Besnard, A., Bernatchez, L., 2018.
Demographic and genetic approaches to study dispersal in wild animal populations: A
methodological review. Mol. Ecol. 27, 3976–4010. https://doi.org/10.1111/mec.14848
Chapman, B.B., Brönmark, C., Nilsson, J.-Å., Hansson, L.-A., 2011. The ecology and evolution of partial
migration. Oikos 120, 1764–1775. https://doi.org/10.1111/j.1600-0706.2011.20131.x
Charnov, E.L., 1976. Optimal foraging, the marginal value theorem. Theor. Popul. Biol. 9, 129–136.
https://doi.org/10.1016/0040-5809(76)90040-X
Chen, K.-Y., Marschall, E.A., Sovic, M.G., Fries, A.C., Gibbs, H.L., Ludsin, S.A., 2018. assignPOP: An R
package for population assignment using genetic, non-genetic, or integrated data in a machine-
learning framework. Methods Ecol. Evol. 9, 439–446. https://doi.org/10.1111/2041-210X.12897
Cherel, Y., Phillips, R.A., Hobson, K.A., McGill, R., 2006. Stable isotope evidence of diverse species-
specific and individual wintering strategies in seabirds. Biol. Lett. 2, 301–303.
https://doi.org/10.1098/rsbl.2006.0445
Chikaraishi, Y., Naraoka, H., Poulson, S.R., 2004. Hydrogen and carbon isotopic fractionations of lipid
biosynthesis among terrestrial (C3, C4 and CAM) and aquatic plants. Phytochemistry 65, 1369–
1381. https://doi.org/10.1016/j.phytochem.2004.03.036
Christie, M.R., Marine, M.L., Fox, S.E., French, R.A., Blouin, M.S., 2016. A single generation of
domestication heritably alters the expression of hundreds of genes. Nat. Commun. 7, 10676.
https://doi.org/10.1038/ncomms10676
Christie, M.R., Marine, M.L., French, R.A., Waples, R.S., Blouin, M.S., 2012. Effective size of a wild
salmonid population is greatly reduced by hatchery supplementation. Heredity 109, 254–260.
https://doi.org/10.1038/hdy.2012.39
Ciancio, J.E., Pascual, M.A., Botto, F., Amaya-Santi, M., O‘Neal, S., Riva Rossi, C., Iribarne, O., 2008.
Stable isotope profiles of partially migratory salmonid populations in Atlantic rivers of Patagonia.
J. Fish Biol. 72, 1708–1719. https://doi.org/10.1111/j.1095-8649.2008.01846.x
Clapp, D.F., Clark, R.D., Diana, J.S., 1990. Range, activity, and habitat of large, free-ranging brown trout
in a Michigan stream. Trans. Am. Fish. Soc. 119, 1022–1034. https://doi.org/10.1577/1548-
8659(1990)119<1022:RAAHOL>2.3.CO;2
Clegg, S.M., Kelly, J.F., Kimura, M., Smith, T.B., 2003. Combining genetic markers and stable isotopes to
reveal population connectivity and migration patterns in a Neotropical migrant, Wilson‘s warbler
(Wilsonia pusilla). Mol. Ecol. 12, 819–830. https://doi.org/10.1046/j.1365-294X.2003.01757.x
Clobert, J. (Ed.), 2001. Dispersal. Oxford University Press, Oxford ; New York.
Clobert, J., Le Galliard, J.-F., Cote, J., Meylan, S., Massot, M., 2009. Informed dispersal, heterogeneity in
animal dispersal syndromes and the dynamics of spatially structured populations. Ecol. Lett. 12,
197–209. https://doi.org/10.1111/j.1461-0248.2008.01267.x
Cook, B.D., Bunn, S.E., Hughes, J.M., 2007. Molecular genetic and stable isotope signatures reveal
complementary patterns of population connectivity in the regionally vulnerable southern pygmy
perch (Nannoperca australis). Biol. Conserv. 138, 60–72.
https://doi.org/10.1016/j.biocon.2007.04.002
Cooke, S.J., Midwood, J.D., Thiem, J.D., Klimley, P., Lucas, M.C., Thorstad, E.B., Eiler, J., Holbrook, C.,
Ebner, B.C., 2013. Tracking animals in freshwater with electronic tags: past, present and future.
Anim. Biotelemetry 1, 5. https://doi.org/10.1186/2050-3385-1-5
Cornuet, J.M., Piry, S., Luikart, G., Estoup, A., Solignac, M., 1999. New methods employing multilocus
genotypes to select or exclude populations as origins of individuals. Genetics 153, 1989–2000.
228
Cortey, M., Garcia-Marin, J.-L., 2002. Evidence for phylogeographically informative sequence variation in
the mitochondrial control region of Atlantic brown trout. J. Fish Biol. 60, 1058–1063.
https://doi.org/10.1111/j.1095-8649.2002.tb02429.x
Cortey, M., Pla, C., García-Marín, J.-L., 2004. Historical biogeography of Mediterranean trout. Mol.
Phylogenet. Evol. 33, 831–844. https://doi.org/10.1016/j.ympev.2004.08.012
Cote, J., Bestion, E., Jacob, S., Travis, J., Legrand, D., Baguette, M., 2017. Evolution of dispersal strategies
and dispersal syndromes in fragmented landscapes. Ecography 40, 56–73.
https://doi.org/10.1111/ecog.02538
Cote, J., Clobert, J., Brodin, T., Fogarty, S., Sih, A., 2010. Personality-dependent dispersal:
characterization, ontogeny and consequences for spatially structured populations. Philos. Trans. R.
Soc. B Biol. Sci. 365, 4065–4076. https://doi.org/10.1098/rstb.2010.0176
Craine, J.M., Brookshire, E.N.J., Cramer, M.D., Hasselquist, N.J., Koba, K., Marin-Spiotta, E., Wang, L.,
2015. Ecological interpretations of nitrogen isotope ratios of terrestrial plants and soils. Plant Soil
396, 1–26. https://doi.org/10.1007/s11104-015-2542-1
Crawford, K., Mcdonald, R.A., Bearhop, S., 2008. Applications of stable isotope techniques to the ecology
of mammals. Mammal Rev. 38, 87–107. https://doi.org/10.1111/j.1365-2907.2008.00120.x
Crisp, D.T., Hurley, M.A., 1991. Stream channel experiments on downstream movement of recently
emerged trout, Salmo trutta L., and salmon, S. salar L.-11. Effects of constant and changing
velocities and of day and night upon dispersal rate. J. Fish Biol. 39, 363–370.
https://doi.org/10.1111/j.1095-8649.1991.tb04368.x
Cucherousset, J., Ombredane, D., Charles, K., Marchand, F., Baglinière, J.-L., 2005. A continuum of life
history tactics in a brown trout (Salmo trutta) population. Can. J. Fish. Aquat. Sci. 62, 1600–1610.
https://doi.org/10.1139/f05-057
Cuinat, R., 1971. Principaux caractères démographiques observés sur 50 rivières à truites françaises.
Influence de la pente et du calcium. Ann. Hydrobiol. 187–207.
Cuinat, R., Heland, M., 1979. Observations sur la dévalaison d‘alevins de truite commune (Salmo trutta L.)
dans le Lissuraga. Bull. Fr. Piscic. 1–17. https://doi.org/10.1051/kmae:1979005
Cunjak, R.A., Roussel, J.-M., Gray, M.A., Dietrich, J.P., Cartwright, D.F., Munkittrick, K.R., Jardine, T.D.,
2005. Using stable isotope analysis with telemetry or mark-recapture data to identify fish
movement and foraging. Oecologia 144, 636–646. https://doi.org/10.1007/s00442-005-0101-9
Cutler, D.R., Edwards, T.C., Beard, K.H., Cutler, A., Hess, K.T., Gibson, J., Lawler, J.J., 2007. Random
forests for classification in ecology. Ecology 88, 2783–2792. https://doi.org/10.1890/07-0539.1
DeNiro, M.J., Epstein, S., 1978. Influence of diet on the distribution of carbon isotopes in animals.
Geochim. Cosmochim. Acta 42, 495–506. https://doi.org/10.1016/0016-7037(78)90199-0
Dieterman, D.J., Hoxmeier, R.J.H., 2011. Demography of juvenile and adult brown trout in streams of
southeastern Minnesota. Trans. Am. Fish. Soc. 140, 1642–1656.
https://doi.org/10.1080/00028487.2011.641883
Dingemanse, N.J., Both, C., van Noordwijk, A.J., Rutten, A.L., Drent, P.J., 2003. Natal dispersal and
personalities in great tits (Parus major). Proc. R. Soc. B Biol. Sci. 270, 741–747.
https://doi.org/10.1098/rspb.2002.2300
Dingle, H., Drake, V.A., 2007. What is migration? BioScience 57, 113–121.
https://doi.org/10.1641/B570206
Douglas, T.A., Chamberlain, C.P., Blum, J.D., 2002. Land use and geologic controls on the major
elemental and isotopic (δ15N and 87Sr/86Sr) geochemistry of the Connecticut River watershed,
USA. Chem. Geol. 189, 19–34. https://doi.org/10.1016/S0009-2541(02)00047-5
Dudgeon, D., Arthington, A.H., Gessner, M.O., Kawabata, Z.-I., Knowler, D.J., Lévêque, C., Naiman, R.J.,
Prieur-Richard, A.-H., Soto, D., Stiassny, M.L.J., Sullivan, C.A., 2006. Freshwater biodiversity:
importance, threats, status and conservation challenges. Biol. Rev. 81, 163.
https://doi.org/10.1017/S1464793105006950
229
Durante, C., Bertacchini, L., Bontempo, L., Camin, F., Manzini, D., Lambertini, P., Marchetti, A., Paolini,
M., 2016. From soil to grape and wine: Variation of light and heavy elements isotope ratios. Food
Chem. 210, 648–659. https://doi.org/10.1016/j.foodchem.2016.04.108
Durbec, M., Cavalli, L., Grey, J., Chappaz, R., Nguyen The, B., 2010. The use of stable isotopes to trace
small-scale movements by small fish species. Hydrobiologia 641, 23–31.
https://doi.org/10.1007/s10750-009-0051-z
Edelaar, P., Bolnick, D.I., 2012. Non-random gene flow: an underappreciated force in evolution and
ecology. Trends Ecol. Evol. 27, 659–665. https://doi.org/10.1016/j.tree.2012.07.009
Einum, S., Sundt-Hansen, L., H. Nislow, K., 2006. The partitioning of density-dependent dispersal, growth
and survival throughout ontogeny in a highly fecund organism. Oikos 113, 489–496.
https://doi.org/10.1111/j.2006.0030-1299.14806.x
Elliott, J.M., 1994. Quantitative ecology and the brown trout, Oxford series in ecology and evolution.
Oxford University Press, Oxford ; New York.
Erman, N.A., 1981. Terrestrial feeding migration and life history of the stream-dwelling caddisfly,
Desmona bethula (Trichoptera: Limnephilidae). Can. J. Zool. 59, 1658–1665.
https://doi.org/10.1139/z81-230
Fagan, W.F., Holmes, E.E., 2005. Quantifying the extinction vortex. Ecol. Lett. 0, 051109031307004.
https://doi.org/10.1111/j.1461-0248.2005.00845.x
Fahrig, L., Merriam, G., 1985. Habitat patch connectivity and population survival: Ecological Archives
E066-008. Ecology 66, 1762–1768. https://doi.org/10.2307/2937372
Faulks, L.K., Gilligan, D.M., Beheregaray, L.B., 2011. The role of anthropogenic vs. natural in-stream
structures in determining connectivity and genetic diversity in an endangered freshwater fish,
Macquarie perch (Macquaria australasica): Anthropogenic vs. natural in-stream structures in M.
australasica. Evol. Appl. 4, 589–601. https://doi.org/10.1111/j.1752-4571.2011.00183.x
Ferguson, A., 2007. Genetics of sea trout, with particular reference to Britain and Ireland, in: Harris, G.,
Milner, N. (Eds.), Sea Trout: Biology, conservation and management. Blackwell Publishing Ltd,
Oxford, UK, pp. 155–182. https://doi.org/10.1002/9780470996027.ch12
Ferguson, A., 1989. Genetic differences among brown trout, Salmo trutta, stocks and their importance for
the conservation and management of the species. Freshw. Biol. 21, 35–46.
https://doi.org/10.1111/j.1365-2427.1989.tb01346.x
Finn, D.S., Bonada, N., Múrria, C., Hughes, J.M., 2011. Small but mighty: headwaters are vital to stream
network biodiversity at two levels of organization. J. North Am. Benthol. Soc. 30, 963–980.
https://doi.org/10.1899/11-012.1
Fischer, J., Lindenmayer, D.B., 2007. Landscape modification and habitat fragmentation: a synthesis. Glob.
Ecol. Biogeogr. 16, 265–280. https://doi.org/10.1111/j.1466-8238.2007.00287.x
Forseth, T., Nesje, T.F., Jonsson, B., Harsaker, K., 1999. Juvenile migration in brown trout: a consequence
of energetic state. J. Anim. Ecol. 68, 783–793. https://doi.org/10.1046/j.1365-2656.1999.00329.x
Fourcade, Y., Chaput-Bardy, A., Secondi, J., Fleurant, C., Lemaire, C., 2013. Is local selection so
widespread in river organisms? Fractal geometry of river networks leads to high bias in outlier
detection. Mol. Ecol. 22, 2065–2073. https://doi.org/10.1111/mec.12158
Frankham, R., Hemmer, H., Ryder, O.A., Cothran, E.G., Soulé, M.E., Murray, N.D., Snyder, M., 1986.
Selection in captive populations. Zoo Biol. 5, 127–138. https://doi.org/10.1002/zoo.1430050207
Franssen, N.R., Gilbert, E.I., James, A.P., Davis, J.E., 2017. Isotopic tissue turnover and discrimination
factors following a laboratory diet switch in Colorado pikeminnow (Ptychocheilus lucius). Can. J.
Fish. Aquat. Sci. 74, 265–272. https://doi.org/10.1139/cjfas-2015-0531
Fry, B., Baltz, D.M., Benfield, M.C., Fleeger, J.W., Gace, A., Haas, H.L., Quiñones-Rivera, Z.J., 2003.
Stable isotope indicators of movement and residency for brown shrimp (Farfantepenaeus aztecus)
in coastal Louisiana marshscapes. Estuaries 26, 82–97. https://doi.org/10.1007/BF02691696
Fryxell, J.M., Sinclair, A.R.E., 1988. Causes and consequences of migration by large herbivores. Trends
Ecol. Evol. 3, 237–241. https://doi.org/10.1016/0169-5347(88)90166-8
230
Gachot-Neveu, H., Lefevre, P., Roeder, J.-J., Henry, C., Poulle, M.-L., 2009. Genetic detection of sex-
biased and age-biased dispersal in a population of wild carnivore, the red fox, Vulpes vulpes.
Zoolog. Sci. 26, 145–152. https://doi.org/10.2108/zsj.26.145
García-Vega, A., Sanz-Ronda, F.J., Fuentes-Pérez, J.F., 2017. Seasonal and daily upstream movements of
brown trout Salmo trutta in an Iberian regulated river. Knowl. Manag. Aquat. Ecosyst. 9.
https://doi.org/10.1051/kmae/2016041
Gascon, C., Lovejoy, T.E., Bierregaard Jr., R.O., Malcolm, J.R., Stouffer, P.C., Vasconcelos, H.L.,
Laurance, W.F., Zimmerman, B., Tocher, M., Borges, S., 1999. Matrix habitat and species richness
in tropical forest remnants. Biol. Conserv. 91, 223–229. https://doi.org/10.1016/S0006-
3207(99)00080-4
Gatehouse, A.G., Woodrow, K.P., 1987. Simultaneous monitoring of flight and oviposition of individual
velvetbean caterpillar moths (by Wales, Barfield & Leppla, 1985): a critique. Physiol. Entomol. 12,
117–121. https://doi.org/10.1111/j.1365-3032.1987.tb00730.x
Giles, P.T., Burgoyne, J.M., 2008. Skole, D.L. and Tucker, C.J. 1993: Tropical deforestation and habitat
fragmentation in the Amazon: satellite data from 1978 to 1988. Science 260, 1905—1910. Prog.
Phys. Geogr. Earth Environ. 32, 575–580. https://doi.org/10.1177/0309133308096755
Gómez-Díaz, E., González-Solis, J., 2007. Geographic assignment of seabirds to their origin: combining
morphologic, genetic, and biogeochemical analyses. Ecol. Appl. Publ. Ecol. Soc. Am. 17, 1484–
1498.
Gosset, C., Rives, J., Labonne, J., 2006. Effect of habitat fragmentation on spawning migration of brown
trout (Salmo trutta L.). Ecol. Freshw. Fish 15, 247–254. https://doi.org/10.1111/j.1600-
0633.2006.00144.x
Gotelli, N.J., Taylor, C.M., 1999. Testing macroecology models with stream-fish assemblages. Evol. Ecol.
Res. 847–858.
Gouraud, V., Capra, H., Sabaton, C., Tissot, L., Lim, P., Vandewalle, F., Fahrner, G., Souchon, Y., 2008.
Long-term simulations of the dynamics of trout populations on river reaches bypassed by
hydroelectric installations-analysis of the impact of different hydrological scenarios. River Res.
Appl. 24, 1185–1205. https://doi.org/10.1002/rra.1129
Gowan, C., Fausch, K.D., 1996. Mobile brook trout in two high-elevation Colorado streams: reevaluating
the concept of restricted movement. Can. J. Fish. Aquat. Sci. 53, 1370–1381.
https://doi.org/10.1139/cjfas-53-6-1370
Gowaty, P.A., 1993. Differential dispersal, local resource competition, and sex ratio variation in birds. Am.
Nat. 141, 263–280. https://doi.org/10.1086/285472
Greenberg, L.A., Giller, P.S., 2001. Individual variation in habitat use and growth of male and female
brown trout. Ecography 24, 212–224. https://doi.org/10.1034/j.1600-0587.2001.240212.x
Greenberg, R., 1989. Neophobia, aversion to open space, and ecological plasticity in Song and Swamp
sparrows. Can. J. Zool. 67, 1194–1199. https://doi.org/10.1139/z89-172
Greenwood, P.J., 1980. Mating systems, philopatry and dispersal in birds and mammals. Anim. Behav. 28,
1140–1162. https://doi.org/10.1016/S0003-3472(80)80103-5
Grevstad, F.S., Klepetka, B.W., 1992. The influence of plant architecture on the foraging efficiencies of a
suite of ladybird beetles feeding on aphids. Oecologia 92, 399–404.
https://doi.org/10.1007/BF00317466
Griesser, M., Halvarsson, P., Sahlman, T., Ekman, J., 2014. What are the strengths and limitations of direct
and indirect assessment of dispersal? Insights from a long-term field study in a group-living bird
species. Behav. Ecol. Sociobiol. 68, 485–497. https://doi.org/10.1007/s00265-013-1663-x
Groom, M.J., Meffe, G.K., Carroll, C.R. (Eds.), 2006. Principles of conservation biology, 3. ed. ed.
Sinauer, Sunderland, Mass.
Guinand, B., 2002. Comparisons of likelihood and machine learning methods of individual classification. J.
Hered. 93, 260–269. https://doi.org/10.1093/jhered/93.4.260
231
Haas, H.L., Freeman, C.J., Logan, J.M., Deegan, L., Gaines, E.F., 2009. Examining mummichog growth
and movement: Are some individuals making intra-season migrations to optimize growth? J. Exp.
Mar. Biol. Ecol. 369, 8–16. https://doi.org/10.1016/j.jembe.2008.09.027
Haddad, N.M., Brudvig, L.A., Clobert, J., Davies, K.F., Gonzalez, A., Holt, R.D., Lovejoy, T.E., Sexton,
J.O., Austin, M.P., Collins, C.D., Cook, W.M., Damschen, E.I., Ewers, R.M., Foster, B.L., Jenkins,
C.N., King, A.J., Laurance, W.F., Levey, D.J., Margules, C.R., Melbourne, B.A., Nicholls, A.O.,
Orrock, J.L., Song, D.-X., Townshend, J.R., 2015. Habitat fragmentation and its lasting impact on
Earth‘s ecosystems. Sci. Adv. 1, e1500052. https://doi.org/10.1126/sciadv.1500052
Hall, L.A., Beissinger, S.R., 2017. Inferring the timing of long-distance dispersal between Rail
metapopulations using genetic and isotopic assignments. Ecol. Appl. 27, 208–218.
https://doi.org/10.1002/eap.1432
Hamilton, K.E., Ferguson, A., Taggart, J.B., Tomasson, T., Walker, A., Fahy, E., 1989. Post-glacial
colonization of brown trout, Salmo trutta L.: Ldh-5 as a phylogeographic marker locus. J. Fish
Biol. 35, 651–664. https://doi.org/10.1111/j.1095-8649.1989.tb03017.x
Hamrick, J.L., Loveless, M.D., 1986. The influence of seed dispersal mechanisms on the genetic structure
of plant populations, in: Estrada, A., Fleming, T.H. (Eds.), frugivores and seed dispersal. Springer
Netherlands, Dordrecht, pp. 211–223. https://doi.org/10.1007/978-94-009-4812-9_20
Hänfling, B., Weetman, D., 2006. Concordant genetic estimators of migration reveal anthropogenically
enhanced source-sink population structure in the river sculpin, Cottus gobio. Genetics 173, 1487–
1501. https://doi.org/10.1534/genetics.105.054296
Hansen, M.M., 2002. Estimating the long-term effects of stocking domesticated trout into wild brown trout
(Salmo trutta) populations: an approach using microsatellite DNA analysis of historical and
contemporary samples. Mol. Ecol. 11, 1003–1015. https://doi.org/10.1046/j.1365-
294X.2002.01495.x
Hanski, I., 1998. Metapopulation dynamics. Nature 396, 41–49. https://doi.org/10.1038/23876
Harrod, C., Grey, J., McCarthy, T.Kieran., Morrissey, M., 2005. Stable isotope analyses provide new
insights into ecological plasticity in a mixohaline population of European eel. Oecologia 144, 673–
683. https://doi.org/10.1007/s00442-005-0161-x
Hastie, T., Tibshirani, R., Friedman, J., 2009. The elements of statistical learning, Springer series in
statistics. Springer New York, New York, NY. https://doi.org/10.1007/978-0-387-84858-7
Hayes, J.M., 2001. Fractionation of carbon and hydrogen isotopes in biosynthetic processes. Rev. Mineral.
Geochem. 43, 225–277. https://doi.org/10.2138/gsrmg.43.1.225
Heggenes, J., Bagliniere, J.L., Cunjak, R.A., 1999. Spatial niche variability for young Atlantic salmon
(Salmo salar) and brown trout (S. trutta) in heterogeneous streams. Ecol. Freshw. Fish 8, 1–21.
https://doi.org/10.1111/j.1600-0633.1999.tb00048.x
Heggenes, J., Traaen, T., 1988. Downstream migration and critical water velocities in stream channels for
fry of four salmonid species. J. Fish Biol. 32, 717–727. https://doi.org/10.1111/j.1095-
8649.1988.tb05412.x
Heland, M., 1980. La dévalaison des alevins de truite commune, Salmo trutta L. II. Activité des alevins
―dévalants‖ comparés aux sédentaires. Ann. Limnol. 16, 247–254.
https://doi.org/10.1051/limn/1980017
Hershey, A.E., Pastor, J., Peterson, B.J., Kling, G.W., 1993. Stable isotopes resolve the drift paradox for
baetis mayflies in an Arctic river. Ecology 74, 2315–2325. https://doi.org/10.2307/1939584
Hindar, K., Jonsson, B., Ryman, N., Ståhl, G., 1991. Genetic relationships among landlocked, resident, and
anadromous brown trout, Salmo trutta L. Heredity 66, 83–91. https://doi.org/10.1038/hdy.1991.11
Hitt, N.P., Angermeier, P.L., 2008. Evidence for fish dispersal from spatial analysis of stream network
topology. J. North Am. Benthol. Soc. 27, 304–320. https://doi.org/10.1899/07-096.1
Hobson, K.A., 1999. Tracing origins and migration of wildlife using stable isotopes: a review. Oecologia
120, 314–326. https://doi.org/10.1007/s004420050865
232
Hobson, K.A., Clark, R.G., 1992. Assessing avian diets using stable isotopes II: factors influencing diet-
tissue fractionation. The Condor 94, 189–197. https://doi.org/10.2307/1368808
Hobson, K.A., Wassenaar, L.I., 2008. Tracking animal migration with stable isotopes. Academic Press,
Amsterdam.
Hobson, K.A., Wassenaar, L.I., 1997. Linking breeding and wintering grounds of neotropical migrant
songbirds using stable hydrogen isotopic analysis of feathers. Oecologia 109, 142–148.
https://doi.org/10.1007/s004420050068
Höjesjö, J., Økland, F., Sundström, L.F., Pettersson, J., Johnsson, J.I., 2007. Movement and home range in
relation to dominance; a telemetry study on brown trout Salmo trutta. J. Fish Biol. 70, 257–268.
https://doi.org/10.1111/j.1095-8649.2006.01299.x
Holden, C., 2006. Inching Toward Movement Ecology. Science 313, 779–782.
https://doi.org/10.1126/science.313.5788.779
Holmes, R., Hayes, J.W., Jiang, W., Quarterman, A., Davey, L.N., 2014. Emigration and mortality of
juvenile brown trout in a New Zealand headwater tributary. Ecol. Freshw. Fish 23, 631–643.
https://doi.org/10.1111/eff.12118
Hutchings, J.A., Gerber, L., 2002. Sex–biased dispersal in a salmonid fish. Proc. R. Soc. Lond. B Biol. Sci.
269, 2487–2493. https://doi.org/10.1098/rspb.2002.2176
Ims, R.A., 1995. Movement patterns related to spatial structures, in: Hansson, L., Fahrig, L., Merriam, G.
(Eds.), Mosaic Landscapes and Ecological Processes. Springer Netherlands, Dordrecht, pp. 85–109.
https://doi.org/10.1007/978-94-011-0717-4_4
Jackson, S.T., Sax, D.F., 2010. Balancing biodiversity in a changing environment: extinction debt,
immigration credit and species turnover. Trends Ecol. Evol. 25, 153–160.
https://doi.org/10.1016/j.tree.2009.10.001
Jiang, M., Zhu, J., Chen, D., Zhang, R., Qiao, G., 2001. Carbon and strontium isotope variations and
responses to sea-level fluctuations in the Ordovician of the Tarim Basin. Sci. China Ser. Earth Sci.
44, 816–823. https://doi.org/10.1007/BF02907094
Johnson, C.G., 1969. Migration and dispersal of insects by flight. Q. J. R. Meteorol. Soc. 96, 154–154.
https://doi.org/10.1002/qj.49709640721
Johnson, C.J., Parker, K.L., Heard, D.C., Gillingham, M.P., 2002. Movement parameters of ungulates and
scale-specific responses to the environment. J. Anim. Ecol. 71, 225–235.
https://doi.org/10.1046/j.1365-2656.2002.00595.x
Johnsson, J.I., Nobbelin, F., Bohlin, T., 1999. Territorial competition among wild brown trout fry: effects
of ownership and body size. J. Fish Biol. 54, 469–472. https://doi.org/10.1111/j.1095-
8649.1999.tb00846.x
Jombart, T., 2008. adegenet: a R package for the multivariate analysis of genetic markers. Bioinformatics
24, 1403–1405. https://doi.org/10.1093/bioinformatics/btn129
Jombart, T., Devillard, S., Balloux, F., 2010. Discriminant analysis of principal components: a new method
for the analysis of genetically structured populations. BMC Genet. 11, 94.
https://doi.org/10.1186/1471-2156-11-94
Jones, R.E., 1977. Movement patterns and egg distribution in cabbage butterflies. J. Anim. Ecol. 46, 195.
https://doi.org/10.2307/3956
Jonsson, B., Jonsson, N., Hansen, L.P., 1991. Differences in life history and migratory behaviour between
wild and hatchery-reared Atlantic salmon in nature. Aquaculture 98, 69–78.
https://doi.org/10.1016/0044-8486(91)90372-E
Kaplan, 1975. Stable isotopes as a guide to biogeochemical processes. Proc. R. Soc. Lond. B Biol. Sci. 189,
183–211. https://doi.org/10.1098/rspb.1975.0052
Karlsson, S., Moen, T., Lien, S., Glover, K.A., Hindar, K., 2011. Generic genetic differences between
farmed and wild Atlantic salmon identified from a 7K SNP-chip: SNP genotyping and applications.
Mol. Ecol. Resour. 11, 247–253. https://doi.org/10.1111/j.1755-0998.2010.02959.x
233
Kasuya, E., 2000. Kin-biased dispersal behaviour in the mango shield scale, Milviscutulus mangiferae.
Anim. Behav. 59, 629–632. https://doi.org/10.1006/anbe.1999.1334
Kawecki, T.J., Holt, R.D., 2002. Evolutionary Consequences of Asymmetric Dispersal Rates. Am. Nat.
160, 333–347. https://doi.org/10.1086/341519
Keefer, M.L., Caudill, C.C., 2014. Homing and straying by anadromous salmonids: a review of
mechanisms and rates. Rev. Fish Biol. Fish. 24, 333–368. https://doi.org/10.1007/s11160-013-
9334-6
Kelly, J.F., 2000. Stable isotopes of carbon and nitrogen in the study of avian and mammalian trophic
ecology. Can. J. Zool. 78, 1–27. https://doi.org/10.1139/z99-165
Kelly, J.F., Ruegg, K.C., Smith, T.B., 2005. Combining isotopic and genetic markers to identify breeding
origins of migrant birds. Ecol. Appl. 15, 1487–1494. https://doi.org/10.1890/04-1704
Kelly-Quinn, M., Bracken, J.J., 1989. Survival of stocked hatchery-reared brown trout, Salmo trutta L., fry
in relation to the carrying capacity of a trout nursery stream. Aquac. Res. 20, 211–226.
https://doi.org/10.1111/j.1365-2109.1989.tb00346.x
Kennedy, B.P., Klaue, A., Blum, J.D., Folt, C.L., Nislow, K.H., 2002. Reconstructing the lives of fish using
Sr isotopes in otoliths. Can. J. Fish. Aquat. Sci. 59, 925–929. https://doi.org/10.1139/f02-070
Kindlmann, P., Burel, F., 2008. Connectivity measures: a review. Landsc. Ecol.
https://doi.org/10.1007/s10980-008-9245-4
Klemetsen, A., Amundsen, P.-A., Dempson, J.B., Jonsson, B., Jonsson, N., O‘Connell, M.F., Mortensen,
E., 2003. Atlantic salmon Salmo salar L., brown trout Salmo trutta L. and Arctic charr Salvelinus
alpinus L.: a review of aspects of their life histories. Ecol. Freshw. Fish 12, 1–59.
https://doi.org/10.1034/j.1600-0633.2003.00010.x
Kneitel, J.M., Chase, J.M., 2004. Trade-offs in community ecology: linking spatial scales and species
coexistence. Ecol. Lett. 7, 69–80. https://doi.org/10.1046/j.1461-0248.2003.00551.x
Knouft, J.H., Spotila, J.R., 2002. Assessment of movements of resident stream brown trout, Salmo trutta L.,
among contiguous sections of stream. Ecol. Freshw. Fish 11, 85–92. https://doi.org/10.1034/j.1600-
0633.2002.110203.x
Koehn, J.D., Crook, D.A., 2013. Movements and migration.
Koenig, W.D., Van Vuren, D., Hooge, P.N., 1996. Detectability, philopatry, and the distribution of
dispersal distances in vertebrates. Trends Ecol. Evol. 11, 514–517. https://doi.org/10.1016/S0169-
5347(96)20074-6
Kokko, H., 2006. From individual dispersal to species ranges: perspectives for a changing world. Science
313, 789–791. https://doi.org/10.1126/science.1128566
Kotsiantis, S.B., 2007. Supervised machine learning: A review of classification techniques. Informatica
249–268.
Krieg, F., Guyomard, R., 1985. Population genetics of French brown trout (Salmo trutta L): large
geographical differentiation of wild populations and high similarity of domesticated stocks. Genet.
Sel. Evol. 17, 225. https://doi.org/10.1186/1297-9686-17-2-225
Labonne, J., Ravigné, V., Parisi, B., Gaucherel, C., 2008. Linking dendritic network structures to
population demogenetics: The downside of connectivity. Oikos 117, 1479–1490.
https://doi.org/10.1111/j.0030-1299.2008.16976.x
Laikre, L., Antunes, A., Concerted Action on Identification, M. and E. ofGenetic R. in the B.T. (Salmo
trutta) (projekt), 1999. Conservation genetic management of brown trout (Salmo trutta) in Europe.
Danmarks Fiskeriundersøgelser, Afd. for Ferskvandsfiskeri, Silkeborg.
Landguth, E.L., Cushman, S.A., Schwartz, M.K., McKelvey, K.S., Murphy, M., Luikart, G., 2010.
Quantifying the lag time to detect barriers in landscape genetics: quantifying the lag time to detect
barriers in landscape genetics. Mol. Ecol. 19, 4179–4191. https://doi.org/10.1111/j.1365-
294X.2010.04808.x
234
Largiader, C.R., Scholl, A., 1996. Genetic introgression between native and introduced brown trout Salmo
trutta L. populations in the Rhone River Basin. Mol. Ecol. 5, 417–426.
https://doi.org/10.1046/j.1365-294X.1996.00099.x
Larinier, M., Travade, F., 2002. Downstream migration: problems and facilities. Bull. Fr. Pêche Piscic.
181–207. https://doi.org/10.1051/kmae/2002102
Larson, W.A., Seeb, J.E., Pascal, C.E., Templin, W.D., Seeb, L.W., 2014. Single-nucleotide
polymorphisms (SNPs) identified through genotyping-by-sequencing improve genetic stock
identification of Chinook salmon (Oncorhynchus tshawytscha) from western Alaska. Can. J. Fish.
Aquat. Sci. 71, 698–708. https://doi.org/10.1139/cjfas-2013-0502
Leitwein, M., Gagnaire, P.-A., Desmarais, E., Guendouz, S., Rohmer, M., Berrebi, P., Guinand, B., 2016.
Genome-wide nucleotide diversity of hatchery-reared Atlantic and Mediterranean strains of brown
trout Salmo trutta compared to wild Mediterranean populations: genome-wide nucleotide diversity
of s. trutta. J. Fish Biol. 89, 2717–2734. https://doi.org/10.1111/jfb.13131
Li, X.-Y., Kokko, H., 2019. Sex-biased dispersal: a review of the theory: Sex-biased dispersal: a review of
the theory. Biol. Rev. 94, 721–736. https://doi.org/10.1111/brv.12475
Linløkken, A., 1993. Efficiency of fishways and impact of dams on the migration of grayling and brown
trout in the Glomma river system, south-eastern Norway. Regul. Rivers Res. Manag. 8, 145–153.
https://doi.org/10.1002/rrr.3450080117
Litvin, S.Y., Weinstein, M.P., 2004. Multivariate analysis of stable-isotope ratios to infer movements and
utilization of estuarine organic matter by juvenile weakfish (Cynoscion regalis). Can. J. Fish.
Aquat. Sci. 61, 1851–1861. https://doi.org/10.1139/f04-121
Lobón-Cervía, J., Sanz, N. (Eds.), 2018. Brown trout: biology, ecology and management. Wiley, Hoboken,
NJ.
Lowe, W.H., McPeek, M.A., 2014. Is dispersal neutral? Trends Ecol. Evol. 29, 444–450.
https://doi.org/10.1016/j.tree.2014.05.009
Lucas, M.C., Baras, E., Thom, T.J., Duncan, A., Slavk, O. (Eds.), 2001. Migration of freshwater fishes.
Blackwell Science Ltd, Oxford, UK. https://doi.org/10.1002/9780470999653
Lundberg, P., 1987. Partial bird migration and evolutionarily stable strategies. J. Theor. Biol. 125, 351–360.
https://doi.org/10.1016/S0022-5193(87)80067-X
Maisse, G., Bagliniere, J.L., 1990. The biology of brown trout, Salmo trutta L., in the River Scorff,
Brittany: a synthesis of studies from 1973 to 1984. Aquac. Res. 21, 95–106.
https://doi.org/10.1111/j.1365-2109.1990.tb00385.x
Manel, S., Gaggiotti, O., Waples, R., 2005. Assignment methods: matching biological questions with
appropriate techniques. Trends Ecol. Evol. 20, 136–142. https://doi.org/10.1016/j.tree.2004.12.004
Martensen, A.C., Pimentel, R.G., Metzger, J.P., 2008. Relative effects of fragment size and connectivity on
bird community in the Atlantic Rain Forest: Implications for conservation. Biol. Conserv. 141,
2184–2192. https://doi.org/10.1016/j.biocon.2008.06.008
Martin, J., Bareille, G., Berail, S., Pecheyran, C., Daverat, F., Bru, N., Tabouret, H., Donard, O., 2013.
Spatial and temporal variations in otolith chemistry and relationships with water chemistry: a useful
tool to distinguish Atlantic salmon Salmo salar parr from different natal streams: Otolith chemistry
of Salmo salar. J. Fish Biol. 82, 1556–1581. https://doi.org/10.1111/jfb.12089
Maruyama, A., Yamada, Y., Rusuwa, B., Yuma, M., 2001. Change in stable nitrogen isotope ratio in the
muscle tissue of a migratory goby, Rhinogobius sp., in a natural setting. Can. J. Fish. Aquat. Sci.
58, 2125–2128. https://doi.org/10.1139/cjfas-58-11-2125
Matsumoto, S., Yamamoto, T., Yamamoto, M., Zavalaga, C.B., Yoda, K., 2017. Sex-related differences in
the foraging movement of streaked shearwaters Calonectris leucomelas Breeding on Awashima
Island in the Sea of Japan. Ornithol. Sci. 16, 23–32. https://doi.org/10.2326/osj.16.23
McCarthy, I.D., Waldron, S., 2000. Identifying migratory Salmo trutta using carbon and nitrogen stable
isotope ratios. Rapid Commun. Mass Spectrom. 14, 1325–1331. https://doi.org/10.1002/1097-
0231(20000815)14:15<1325::AID-RCM980>3.0.CO;2-A
235
McPeek, M.A., Holt, R.D., 1992. The evolution of dispersal in spatially and temporally varying
environments. Am. Nat. 140, 1010–1027. https://doi.org/10.1086/285453
Mehlhorn, K., Newell, B.R., Todd, P.M., Lee, M.D., Morgan, K., Braithwaite, V.A., Hausmann, D.,
Fiedler, K., Gonzalez, C., 2015. Unpacking the exploration–exploitation tradeoff: A synthesis of
human and animal literatures. Decision 2, 191–215. https://doi.org/10.1037/dec0000033
Mehner, T., 2012. Diel vertical migration of freshwater fishes - proximate triggers, ultimate causes and
research perspectives: Diel vertical migration in freshwater fishes. Freshw. Biol. 57, 1342–1359.
https://doi.org/10.1111/j.1365-2427.2012.02811.x
Mercier, J.-L., Lenoir, A., 1999. Individual flexibility and choice of foraging strategy in Polyrhachis
laboriosa F. Smith (Hymenoptera, Formicidae). Insectes Sociaux 46, 267–272.
https://doi.org/10.1007/s000400050145
Metzger, J.-P., Décamps, H., 1997. The structural connectivity threshold: An hypothesis in conservation
biology at the landscape scale. Acta Oecologica 18, 1–12. https://doi.org/10.1016/S1146-
609X(97)80075-6
Mills, D., 1989. Conservation and management of brown trout, Salmo trutta, in Scotland: an historical
review and the future. Freshw. Biol. 21, 87–98. https://doi.org/10.1111/j.1365-
2427.1989.tb01350.x
Morita, K., Tsuboi, J.-I., Matsuda, H., 2004. The impact of exotic trout on native charr in a Japanese
stream: Impacts of exotic trout on native charr. J. Appl. Ecol. 41, 962–972.
https://doi.org/10.1111/j.0021-8901.2004.00927.x
Morrissey, M.B., de Kerckhove, D.T., 2009. The maintenance of genetic variation due to asymmetric gene
flow in dendritic metapopulations. Am. Nat. 174, 875–889. https://doi.org/10.1086/648311
Moyle, P.B., 1986. Fish Introductions into North America: patterns and ecological impact, in: Mooney,
H.A., Drake, J.A. (Eds.), Ecology of Biological Invasions of North America and Hawaii. Springer
New York, New York, NY, pp. 27–43. https://doi.org/10.1007/978-1-4612-4988-7_2
Nathan, R., Getz, W.M., Revilla, E., Holyoak, M., Kadmon, R., Saltz, D., Smouse, P.E., 2008. A movement
ecology paradigm for unifying organismal movement research. Proc. Natl. Acad. Sci. 105, 19052–
19059. https://doi.org/10.1073/pnas.0800375105
Negro, J.J., Hiraldo, F., Donazar, J.A., 1997. Causes of natal dispersal in the lesser kestrel: inbreeding
avoidance or resource competition? J. Anim. Ecol. 66, 640. https://doi.org/10.2307/5917
Newsome, S.D., Clementz, M.T., Koch, P.L., 2010. Using stable isotope biogeochemistry to study marine
mammal ecology. Mar. Mammal Sci. https://doi.org/10.1111/j.1748-7692.2009.00354.x
Ning, J., Beiko, R.G., 2015. Phylogenetic approaches to microbial community classification. Microbiome
3. https://doi.org/10.1186/s40168-015-0114-5
Northcote, T.G., 1992. Migration and residency in stream salmonids-some ecological considerations and
evolutionary consequences. Nord. J. Freshw. Res. 5–17.
O‘Brian, W.J., Browman, H.I., Evans, B.I., 1990. Search strategies of foraging animals. Am. Sci. 152–160.
Olsen, E.M., Vollestad, L.A., 2005. Small-scale spatial variation in age and size at maturity of stream-
dwelling brown trout, Salmo trutta. Ecol. Freshw. Fish 14, 202–208.
https://doi.org/10.1111/j.1600-0633.2005.00094.x
Olsson, I.C., Greenberg, L.A., Bergman, E., Wysujack, K., 2006. Environmentally induced migration: the
importance of food. Ecol. Lett. 9, 645–651. https://doi.org/10.1111/j.1461-0248.2006.00909.x
Ovidio, M., Baras, E., Goffaux, D., Birtles, C., Philippart, J.C., 1998. Environmental unpredictability rules
the autumn migration of brown trout (Salmo trutta L.) in the Belgian Ardennes, in: Lagardère, J.-P.,
Anras, M.-L.B., Claireaux, G. (Eds.), Advances in Invertebrates and Fish Telemetry. Springer
Netherlands, Dordrecht, pp. 263–274. https://doi.org/10.1007/978-94-011-5090-3_30
Paetkau, D., Calvert, W., Stirling, I., Strobeck, C., 1995. Microsatellite analysis of population structure in
Canadian polar bears. Mol. Ecol. 4, 347–354. https://doi.org/10.1111/j.1365-294X.1995.tb00227.x
236
Paetkau, D., Slade, R., Burden, M., Estoup, A., 2004. Genetic assignment methods for the direct, real-time
estimation of migration rate: a simulation-based exploration of accuracy and power. Mol. Ecol. 13,
55–65.
Patrick, S.C., Charmantier, A., Weimerskirch, H., 2013. Differences in boldness are repeatable and
heritable in a long-lived marine predator. Ecol. Evol. 3, 4291–4299.
https://doi.org/10.1002/ece3.748
Patrick, S.C., Pinaud, D., Weimerskirch, H., 2017. Boldness predicts an individual‘s position along an
exploration-exploitation foraging trade-off. J. Anim. Ecol. 86, 1257–1268.
https://doi.org/10.1111/1365-2656.12724
Paz-Vinas, I., Blanchet, S., 2015. Dendritic connectivity shapes spatial patterns of genetic diversity: a
simulation-based study. J. Evol. Biol. 28, 986–994. https://doi.org/10.1111/jeb.12626
Paz-Vinas, I., Comte, L., Chevalier, M., Dubut, V., Veyssiere, C., Grenouillet, G., Loot, G., Blanchet, S.,
2013. Combining genetic and demographic data for prioritizing conservation actions: insights from
a threatened fish species. Ecol. Evol. 3, 2696–2710. https://doi.org/10.1002/ece3.645
Paz-Vinas, I., Loot, G., Stevens, V.M., Blanchet, S., 2015. Evolutionary processes driving spatial patterns
of intraspecific genetic diversity in river ecosystems. Mol. Ecol. 24, 4586–4604.
https://doi.org/10.1111/mec.13345
Pelletier, F., Garant, D., Hendry, A.P., 2009. Eco-evolutionary dynamics. Philos. Trans. R. Soc. B Biol.
Sci. 364, 1483–1489. https://doi.org/10.1098/rstb.2009.0027
Péron, C., Grémillet, D., 2013. Tracking through life stages: adult, immature and juvenile autumn migration
in a long-lived seabird. PLoS ONE 8, e72713. https://doi.org/10.1371/journal.pone.0072713
Perrier, C., Guyomard, R., Bagliniere, J.-L., Nikolic, N., Evanno, G., 2013. Changes in the genetic structure
of Atlantic salmon populations over four decades reveal substantial impacts of stocking and
potential resiliency. Ecol. Evol. 3, 2334–2349. https://doi.org/10.1002/ece3.629
Perrin, N., Mazalov, V., 1999. Dispersal and inbreeding avoidance. Am. Nat. 154, 282–292.
https://doi.org/10.1086/303236
Peterson, B.J., Fry, B., 1987. Stable isotopes in ecosystem studies. Annu. Rev. Ecol. Syst. 18, 293–320.
https://doi.org/10.1146/annurev.es.18.110187.001453
Pirhonen, J., Forsman, L., 1998. Effect of prolonged feed restriction on size variation, feed consumption,
body composition, growth and smolting of brown trout, Salmo trutta. Aquaculture 162, 203–217.
https://doi.org/10.1016/S0044-8486(98)00215-4
Plath, M., Pfenninger, M., Lerp, H., Riesch, R., Eschenbrenner, C., Slattery, P.A., Bierbach, D., Herrmann,
N., Schulte, M., Arias-Rodriguez, L., Rimber Indy, J., Passow, C.N., Tobler, M., 2013. Data from:
Genetic differentiation and selection against migrants in evolutionarily replicated extreme
environments. https://doi.org/10.5061/dryad.7cn23
Pollux, B.J.A., Luteijn, A., Van Groenendael, J.M., Ouborg, N.J., 2009. Gene flow and genetic structure of
the aquatic macrophyte Sparganium emersum in a linear unidirectional river. Freshw. Biol. 54, 64–
76. https://doi.org/10.1111/j.1365-2427.2008.02100.x
Post, D.M., Palkovacs, E.P., Schielke, E.G., Dodson, S.I., 2008. Intraspecific variation in a predator affects
community structure and cascading trophic interactions. Ecology 89, 2019–2032.
https://doi.org/10.1890/07-1216.1
Postlethwaite, C.M., Brown, P., Dennis, T.E., 2013. A new multi-scale measure for analysing animal
movement data. J. Theor. Biol. 317, 175–185. https://doi.org/10.1016/j.jtbi.2012.10.007
Pritchard, J.K., Stephens, M., Donnelly, P., 2000. Inference of population structure using multilocus
genotype data. Genetics 155, 945–959.
Prunier, J.G., Dubut, V., Loot, G., Tudesque, L., Blanchet, S., 2018. The relative contribution of river
network structure and anthropogenic stressors to spatial patterns of genetic diversity in two
freshwater fishes: A multiple-stressors approach. Freshw. Biol. 63, 6–21.
https://doi.org/10.1111/fwb.13034
237
Prunier, J.G., Kaufmann, B., Léna, J.-P., Fenet, S., Pompanon, F., Joly, P., 2014. A 40-year-old divided
highway does not prevent gene flow in the alpine newt Ichthyosaura alpestris. Conserv. Genet. 15,
453–468. https://doi.org/10.1007/s10592-013-0553-0
Pulido, F., Berthold, P., 2003. Quantitative genetic analysis of migratory behaviour, in: Berthold, P.,
Gwinner, E., Sonnenschein, E. (Eds.), Avian Migration. Springer Berlin Heidelberg, Berlin,
Heidelberg, pp. 53–77. https://doi.org/10.1007/978-3-662-05957-9_4
Quinn, T.P., 1984. Homing and straying in Pacific salmon, in: McCleave, J.D., Arnold, G.P., Dodson, J.J.,
Neill, W.H. (Eds.), Mechanisms of migration in fishes. Springer US, Boston, MA, pp. 357–362.
https://doi.org/10.1007/978-1-4613-2763-9_21
Radinger, J., Wolter, C., 2014. Patterns and predictors of fish dispersal in rivers. Fish Fish. 15, 456–473.
https://doi.org/10.1111/faf.12028
Raeymaekers, J.A.M., Maes, G.E., Geldof, S., Hontis, I., Nackaerts, K., Volckaert, F.A.M., 2008. Modeling
genetic connectivity in sticklebacks as a guideline for river restoration. Evol. Appl. 1, 475–488.
https://doi.org/10.1111/j.1752-4571.2008.00019.x
Rahel, F.J., Nibbelink, N.P., 1999. Spatial patterns in relations among brown trout (Salmo trutta)
distribution, summer air temperature, and stream size in Rocky Mountain streams. Can. J. Fish.
Aquat. Sci. 56, 43–51. https://doi.org/10.1139/f99-210
Rakow, T., Newell, B.R., Zougkou, K., 2010. The role of working memory in information acquisition and
decision making: lessons from the binary prediction task. Q. J. Exp. Psychol. 63, 1335–1360.
https://doi.org/10.1080/17470210903357945
Rakowitz, G., Berger, B., Kubecka, J., Keckeis, H., 2008. Functional role of environmental stimuli for the
spawning migration in Danube nase Chondrostoma nasus (L.). Ecol. Freshw. Fish 17, 502–514.
https://doi.org/10.1111/j.1600-0633.2008.00302.x
Rannala, B., Mountain, J.L., 1997. Detecting immigration by using multilocus genotypes. Proc. Natl. Acad.
Sci. 94, 9197–9201. https://doi.org/10.1073/pnas.94.17.9197
Rasmussen, J.B., Trudeau, V., Morinville, G., 2009. Estimating the scale of fish feeding movements in
rivers using δ 13
C signature gradients. J. Anim. Ecol. 78, 674–685. https://doi.org/10.1111/j.1365-
2656.2008.01511.x
Reed, J.M., Boulinier, T., Danchin, E., Oring, L.W., 1999. Informed dispersal, in: Nolan, V., Ketterson,
E.D., Thompson, C.F. (Eds.), Current Ornithology. Springer US, Boston, MA, pp. 189–259.
https://doi.org/10.1007/978-1-4757-4901-4_5
Rikardsen, A.H., Hansen, L.P., Jensen, A.J., Vollen, T., Finstad, B., 2008. Do Norwegian Atlantic salmon
feed in the northern Barents Sea? Tag recoveries from 70 to 78° N. J. Fish Biol. 72, 1792–1798.
https://doi.org/10.1111/j.1095-8649.2008.01823.x
Riveros, A.J., Gronenberg, W., 2010. Sensory allometry, foraging task specialization and resource
exploitation in honeybees. Behav. Ecol. Sociobiol. 64, 955–966. https://doi.org/10.1007/s00265-
010-0911-6
Rolshausen, G., Segelbacher, G., Hermes, C., Hobson, K.A., Schaefer, H.M., 2013. Individual differences
in migratory behavior shape population genetic structure and microhabitat choice in sympatric
blackcaps (Sylvia atricapilla). Ecol. Evol. 3, 4278–4289. https://doi.org/10.1002/ece3.825
Romare, P., Hansson, L.-A., 2003. A behavioral cascade: Top-predator induced behavioral shifts in
planktivorous fish and zooplankton. Limnol. Oceanogr. 48, 1956–1964.
https://doi.org/10.4319/lo.2003.48.5.1956
Ronce, O., 2007. How Does it feel to be like a rolling stone? Ten questions about dispersal evolution.
Annu. Rev. Ecol. Evol. Syst. 38, 231–253.
https://doi.org/10.1146/annurev.ecolsys.38.091206.095611
Roussel, J.M., Bardonnet, A., 2002. Habitat de la truite commune (Salmo trutta L.) pendant la période
juvénile en ruisseau : préférences, mouvements, variations journalières et saisonnières. Bull. Fr.
Pêche Piscic. 435–454. https://doi.org/10.1051/kmae:2002044
238
Rubenstein, D.R., Hobson, K.A., 2004. From birds to butterflies: animal movement patterns and stable
isotopes. Trends Ecol. Evol. 19, 256–263. https://doi.org/10.1016/j.tree.2004.03.017
Rundel, C.W., Wunder, M.B., Alvarado, A.H., Ruegg, K.C., Harrigan, R., Schuh, A., Kelly, J.F., Siegel,
R.B., DeSante, D.F., Smith, T.B., Novembre, J., 2013. Novel statistical methods for integrating
genetic and stable isotope data to infer individual-level migratory connectivity. Mol. Ecol. 22,
4163–4176. https://doi.org/10.1111/mec.12393
Runge, C.A., Martin, T.G., Possingham, H.P., Willis, S.G., Fuller, R.A., 2014. Conserving mobile species.
Front. Ecol. Environ. 12, 395–402. https://doi.org/10.1890/130237
Saint-Pé, K., Blanchet, S., Tissot, L., Poulet, N., Plasseraud, O., Loot, G., Veyssière, C., Prunier, J.G.,
2018. Genetic admixture between captive-bred and wild individuals affects patterns of dispersal in
a brown trout (Salmo trutta) population. Conserv. Genet. 19, 1269–1279.
https://doi.org/10.1007/s10592-018-1095-2
Sanz, N., Fernández-Cebrián, R., Casals, F., Araguas, R.M. and García-Marín, J.L., 2011. Dispersal and
demography of brown trout, Salmo trutta, inferred from population and family structure in unstable
Mediterranean streams. Hydrobiologia, 671(1), p.105.
Sanz, N., García-Marín, J.-L., Pla, C., 2000. Divergence of brown trout (Salmo trutta) within glacial
refugia. Can. J. Fish. Aquat. Sci. 57, 2201–2210. https://doi.org/10.1139/f00-199
Schager, E., Peter, A., Burkhardt-Holm, P., 2007. Status of young-of-the-year brown trout (Salmo trutta
fario) in Swiss streams: factors influencing YOY trout recruitment. Aquat. Sci. 69, 41–50.
https://doi.org/10.1007/s00027-007-0845-x
Schell, D.M., Saupe, S.M., Haubenstock, N., 1989. Bowhead whale (Balaena mysticetus) growth and
feeding as estimated by ?13C techniques. Mar. Biol. 103, 433–443.
https://doi.org/10.1007/BF00399575
Schick, R.S., Loarie, S.R., Colchero, F., Best, B.D., Boustany, A., Conde, D.A., Halpin, P.N., Joppa, L.N.,
McClellan, C.M., Clark, J.S., 2008. Understanding movement data and movement processes:
current and emerging directions. Ecol. Lett. 11, 1338–1350. https://doi.org/10.1111/j.1461-
0248.2008.01249.x
Schilt, C.R., 2007. Developing fish passage and protection at hydropower dams. Appl. Anim. Behav. Sci.
104, 295–325. https://doi.org/10.1016/j.applanim.2006.09.004
Schlosser, I.J., 1998. Fish recruitment, dispersal, and trophic interactions in a heterogeneous lotic
environment. Oecologia 113, 260–268. https://doi.org/10.1007/s004420050377
Schnell, J.K., Harris, G.M., Pimm, S.L., Russell, G.J., 2013. Quantitative analysis of forest fragmentation
in the Atlantic forest reveals more threatened bird species than the current red list. PLoS ONE 8,
e65357. https://doi.org/10.1371/journal.pone.0065357
Schrider, D.R., Kern, A.D., 2018. Supervised machine learning for population genetics: a new paradigm.
Trends Genet. 34, 301–312. https://doi.org/10.1016/j.tig.2017.12.005
Sekercioglu, C.H., 2002. Effects of forestry practices on vegetation structure and bird community of Kibale
National Park, Uganda. Biol. Conserv. 107, 229–240. https://doi.org/10.1016/S0006-
3207(02)00097-6
Shaw, A.K., Couzin, I.D., 2013. Migration or residency? The evolution of movement behavior and
information usage in seasonal environments. Am. Nat. 181, 114–124.
https://doi.org/10.1086/668600
Sheehan, S., Song, Y.S., 2016. Deep learning for population genetic inference. PLOS Comput. Biol. 12,
e1004845. https://doi.org/10.1371/journal.pcbi.1004845
Skalski, G.T., Gilliam, J.F., 2000. Modeling diffusive spread in a heterogeneous population: a movement
study with stream fish. Ecology 81, 1685–1700. https://doi.org/10.1890/0012-
9658(2000)081[1685:MDSIAH]2.0.CO;2
Skov, C., Chapman, B.B., Baktoft, H., Brodersen, J., Bronmark, C., Hansson, L.-A., Hulthen, K., Nilsson,
P.A., 2013. Migration confers survival benefits against avian predators for partially migratory
freshwater fish. Biol. Lett. 9, 20121178–20121178. https://doi.org/10.1098/rsbl.2012.1178
239
Slavik, O., Bartos, L., 2004. Brown trout migration and flow variability. Int. J. Ecohydrol. Hydrobiol. 157–
163. Snoj, A., Marić, S., Bajec, S.S., Berrebi, P., Janjani, S. and Schöffmann, J., 2011. Phylogeographic
structure and demographic patterns of brown trout in North-West Africa. Molecular
Phylogenetics and Evolution, 61(1), pp.203-211. Solomon, D.J., Templeton, R.G., 1976. Movements of brown trout Salmo trutta L. in a chalk stream. J. Fish
Biol. 9, 411–423. https://doi.org/10.1111/j.1095-8649.1976.tb04690.x
Splendiani, A., Ruggeri, P., Giovannotti, M., Pesaresi, S., Occhipinti, G., Fioravanti, T., Lorenzoni, M.,
Nisi Cerioni, P., Caputo Barucchi, V., 2016. Alien brown trout invasion of the Italian peninsula: the
role of geological, climate and anthropogenic factors. Biol. Invasions 18, 2029–2044.
https://doi.org/10.1007/s10530-016-1149-7
Stevens, V.M., Verkenne, C., Vandewoestijne, S., Wesselingh, R.A., Baguette, M., 2006. Gene flow and
functional connectivity in the natterjack toad: genetic validation of functional connectivity. Mol.
Ecol. 15, 2333–2344. https://doi.org/10.1111/j.1365-294X.2006.02936.x
Susnik, S., Schoffmann, J., Weiss, S., 2005. Genetic verification of native brown trout from the Persian
Gulf (Catak Cay River, Tigris basin). J. Fish Biol. 67, 879–884. https://doi.org/10.1111/j.0022-
1112.2005.00780.x
Swatdipong, A., Vasemägi, A., Niva, T., Koljonen, M.-L., Primmer, C.R., 2010. High level of population
genetic structuring in lake-run brown trout, Salmo trutta, of the Inari Basin, northern Finland. J.
Fish Biol. 77, 2048–2071. https://doi.org/10.1111/j.1095-8649.2010.02784.x
Sweeney, K., Cusack, B., Armagost, F., O‘Brien, T., Keiser, C.N., Pruitt, J.N., 2013. Predator and prey
activity levels jointly influence the outcome of long-term foraging bouts. Behav. Ecol. 24, 1205–
1210. https://doi.org/10.1093/beheco/art052
Sylvester, E.V.A., Bentzen, P., Bradbury, I.R., Clément, M., Pearce, J., Horne, J., Beiko, R.G., 2018.
Applications of random forest feature selection for fine-scale genetic population assignment. Evol.
Appl. 11, 153–165. https://doi.org/10.1111/eva.12524
Szulkin, M., Sheldon, B.C., 2008. Dispersal as a means of inbreeding avoidance in a wild bird population.
Proc. R. Soc. B Biol. Sci. 275, 703–711. https://doi.org/10.1098/rspb.2007.0989
Taylor, L.R., 1986. Synoptic dynamics, migration and the rothamsted insect survey: presidential address to
the British Ecological Society, December 1984. J. Anim. Ecol. 55, 1. https://doi.org/10.2307/4690
Taylor, P.D., Fahrig, L., Henein, K., Merriam, G., 1993. Connectivity is a vital element of landscape
structure. Oikos 68, 571. https://doi.org/10.2307/3544927
Templin, W.D., Seeb, J.E., Jasper, J.R., Barclay, A.W., Seeb, L.W., 2011. Genetic differentiation of Alaska
Chinook salmon: the missing link for migratory studies: SNP genotyping and applications. Mol.
Ecol. Resour. 11, 226–246. https://doi.org/10.1111/j.1755-0998.2010.02968.x
Thompson, D.R., Furness, R.W., 1995. Stable-isotope ratios of Carbon and Nitrogen in feathers indicate
seasonal dietary shifts in northern fulmars. The Auk 112, 493–498.
https://doi.org/10.2307/4088739
Tieszen, L.L., 1991. Natural variations in the carbon isotope values of plants: Implications for archaeology,
ecology, and paleoecology. J. Archaeol. Sci. 18, 227–248. https://doi.org/10.1016/0305-
4403(91)90063-U
Tischendorf, L., Fahrig, L., 2000. On the usage and measurement of landscape connectivity. Oikos 90, 7–
19. https://doi.org/10.1034/j.1600-0706.2000.900102.x
Tomanova, S., Courret, D., Alric, A., De Oliveira, E., Lagarrigue, T., Tétard, S., 2018. Protecting
efficiently sea-migrating salmon smolts from entering hydropower plant turbines with inclined or
oriented low bar spacing racks. Ecol. Eng. 122, 143–152.
https://doi.org/10.1016/j.ecoleng.2018.07.034
Tougard, C., Justy, F., Guinand, B., Douzery, E.J.P., Berrebi, P., 2018. Salmo macrostigma (Teleostei,
Salmonidae): Nothing more than a brown trout (S. trutta) lineage? J. Fish Biol. 93, 302–310.
https://doi.org/10.1111/jfb.13751
240
Touw, W.G., Bayjanov, J.R., Overmars, L., Backus, L., Boekhorst, J., Wels, M., van Hijum, S.A.F.T.,
2013. Data mining in the Life Sciences with Random Forest: a walk in the park or lost in the
jungle? Brief. Bioinform. 14, 315–326. https://doi.org/10.1093/bib/bbs034
Townsend, C.R., 1996. Invasion biology and ecological impacts of brown trout Salmo trutta in New
Zealand. Biol. Conserv. 78, 13–22. https://doi.org/10.1016/0006-3207(96)00014-6
Uezu, A., Metzger, J.P., Vielliard, J.M.E., 2005. Effects of structural and functional connectivity and patch
size on the abundance of seven Atlantic Forest bird species. Biol. Conserv. 123, 507–519.
https://doi.org/10.1016/j.biocon.2005.01.001
Van Dyck, H., Baguette, M., 2005. Dispersal behaviour in fragmented landscapes: Routine or special
movements? Basic Appl. Ecol. 6, 535–545. https://doi.org/10.1016/j.baae.2005.03.005
van Noordwijk, A.J., Pulido, F., Helm, B., Coppack, T., Delingat, J., Dingle, H., Hedenström, A., van der
Jeugd, H., Marchetti, C., Nilsson, A., Pérez-Tris, J., 2006. A framework for the study of genetic
variation in migratory behaviour. J. Ornithol. 147, 221–233. https://doi.org/10.1007/s10336-005-
0047-z
Vannote, R.L., Minshall, G.W., Cummins, K.W., Sedell, J.R., Cushing, C.E., 1980. The River Continuum
Concept. Can. J. Fish. Aquat. Sci. 37, 130–137. https://doi.org/10.1139/f80-017
Vera, M., Cortey, M., Sanz, N., García-Marín, J.-L., 2010. Maintenance of an endemic lineage of brown
trout (Salmo trutta) within the Duero river basin. J. Zool. Syst. Evol. Res. 48, 181–187.
https://doi.org/10.1111/j.1439-0469.2009.00547.x
Vera, M., García-Marín, J.L., Martinez, P., Bouza, C., 2015. Phylogenetic diversity within the endemic
brown trout Duero lineage: implications for conservation and management. Mar. Freshw. Res. 66,
1066. https://doi.org/10.1071/MF14286
Vollestad, L.A., Olsen, E.M., Forseth, T., 2002. Growth-rate variation in brown trout in small neighbouring
streams: evidence for density-dependence? J. Fish Biol. 61, 1513–1527.
https://doi.org/10.1111/j.1095-8649.2002.tb02494.x
Vøllestad, L.A., Serbezov, D., Bass, A., Bernatchez, L., Olsen, E.M., Taugbøl, A., 2012. Small-scale
dispersal and population structure in stream-living brown trout (Salmo trutta) inferred by mark–
recapture, pedigree reconstruction, and population genetics. Can. J. Fish. Aquat. Sci. 69, 1513–
1524. https://doi.org/10.1139/f2012-073
Walther, B., Thorrold, S., 2006. Water, not food, contributes the majority of strontium and barium
deposited in the otoliths of a marine fish. Mar. Ecol. Prog. Ser. 311, 125–130.
https://doi.org/10.3354/meps311125
Ward, R.D., Woodwark, M., Skibinski, D.O.F., 1994. A comparison of genetic diversity levels in marine,
freshwater, and anadromous fishes. J. Fish Biol. 44, 213–232. https://doi.org/10.1111/j.1095-
8649.1994.tb01200.x
Webster, M.S., Marra, P.P., Haig, S.M., Bensch, S., Holmes, R.T., 2002. Links between worlds: unraveling
migratory connectivity. Trends Ecol. Evol. 17, 76–83. https://doi.org/10.1016/S0169-
5347(01)02380-1
Westley, P.A.H., Quinn, T.P., Dittman, A.H., 2013. Rates of straying by hatchery-produced Pacific salmon
(Oncorhynchus spp.) and steelhead (Oncorhynchus mykiss) differ among species, life history types,
and populations. Can. J. Fish. Aquat. Sci. 70, 735–746. https://doi.org/10.1139/cjfas-2012-0536
Whitlock, M.C., McCauley, D.E., 1999. Indirect measures of gene flow and migration: FST not equal to
1/(4Nm + 1). Heredity 82 ( Pt 2), 117–125.
Whoriskey, F., 2005. Evolution illuminated: salmon and their relatives . Edited by Andrew P Hendry and ,
Stephen C Stearns. Oxford and New York: Oxford University Press .510 p; ill.; index. ISBN: 0–
19–514385‐X. 2004. Q. Rev. Biol. 80, 115–116. https://doi.org/10.1086/431065
Williams, M., 2006. Deforesting the earth: from prehistory to global crisis: an abridgment. University of
Chicago Press, Chicago.
Wilson, G.A., Rannala, B., 2003. Bayesian inference of recent migration rates using multilocus genotypes.
Genetics 163, 1177–1191.
241
Wocher, H., Rösch, R., 2006. Migrations of soufie (Leuciscus souffia agassizii, VAL. 1844) in a natural
river and a tributary determined by mark-recapture. Arch. Für Hydrobiol. 165, 77–87.
https://doi.org/10.1127/0003-9136/2006/0165-0077
Wood, C.C., Foote, C.J., 1996. Evidence for sympatric genetic divergence of anadromous and
nonanadromous morphs of Sockeye salmon (Oncorhynchus nerka). Evolution 50, 1265–1279.
https://doi.org/10.1111/j.1558-5646.1996.tb02367.x
Wright, S., 1931. Evolution in mendelian populations. Genetics 16, 97–159.
Young, A., Boyle, T., Brown, T., 1996. The population genetic consequences of habitat fragmentation for
plants. Trends Ecol. Evol. 11, 413–418. https://doi.org/10.1016/0169-5347(96)10045-8
Zera, A.J., Denno, R.F., 1997. Physiology and ecology of dispersal polymorphism in insects. Annu. Rev.
Entomol. 42, 207–230. https://doi.org/10.1146/annurev.ento.42.1.207
Ziółkowska, E., Ostapowicz, K., Radeloff, V.C., Kuemmerle, T., Sergiel, A., Zwijacz-Kozica, T., Zięba, F.,
Śmietana, W., Selva, N., 2016. Assessing differences in connectivity based on habitat versus
movement models for brown bears in the Carpathians. Landsc. Ecol. 31, 1863–1882.
https://doi.org/10.1007/s10980-016-0368-8